For Peer Review Only Journal: BMJ Open

BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from

Fetal intracranial haemorrhages caused by fetal and neonatal alloimmune thrombocytopenia: An observational cohort study of 43 cases from an international multicentre registry

For peer review only Journal: BMJ Open

Manuscript ID: bmjopen-2012-002490

Article Type: Research

Date Submitted by the Author: 13-Dec-2012

Complete List of Authors: Tiller, Heidi; University Hospital North Norway, Obstetrics and Gynaecology Kamphuis, Marije; Leiden University Medical Center, Obstetrics Flodmark, Olof; Karolinska Universitetssjukhuset, Neuroradiologiska kliniken Papadogiannakis, Nikos; Karolinska Institutet, Laboratory Medicine David, Anna; University College London Medical School, Institute for Womens Health Sainio, Susanna; Finnish Red Cross Blood Service, Koskinen, Sinikka; Finnish Red Cross Blood Service, Javela, Kaija; Finnish Red Cross Blood Service, Wikman, Agneta; Karolinska University Hospital, Clinical Immunology and Transfusion Medicine

Kekomaki, Riitta; Finnish Red Cross Blood Service, http://bmjopen.bmj.com/ Kanhai, Humphrey; University Medical Center, Obstetrics Gynaecology Oepkes, Dick; Leiden University Medical Center, Obstetrics Husebekk, Anne; University of Tromsø, Institute of Medical Biology Westgren, Magnus; Department of Clinical Science, Intervention and Technology, Karolinska Institute, Department of Clinical Science, Intervention and Technology

Primary Subject Obstetrics and gynaecology

Heading: on October 2, 2021 by guest. Protected copyright.

Secondary Subject Heading: Immunology (including allergy)

Fetal medicine < OBSTETRICS, Prenatal diagnosis < OBSTETRICS, Keywords: Ultrasonography < OBSTETRICS

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Fetal intracranial haemorrhages caused by fetal and neonatal alloimmune 4 5 thrombocytopenia: An observational cohort study of 43 cases from an international 6 multicentre registry 7 1 2 3 8 Heidi Tiller HT, Marije M. Kamphuis MMK, Olof Flodmark OF, Nikos Papadogiannakis 9 4 5 6 6 6 10 NP, Anna L David ALD, Susanna Sainio SS, Sinikka Koskinen SK, Kaija Javela KJ, 11 Agneta Wikman AW,7 Riitta Kekomaki RK,6 Humphrey H.H. Kanhai HHHK,2 Dick Oepkes 12 2 10 11 13 DO, Anne Husebekk AH, * and Magnus Westgren MW * 14 *Corresponding authors 15 For peer review only 16 17 1. Department of Obstetrics and Gynaecology, University Hospital North Norway, 9038 18 Tromsø, Norway 19 20 2. Department of Obstetrics, Leiden University Medical Center, 2253LZ Leiden, The 21 Netherlands 22 3. Neuroradiologiska kliniken, Karolinska Universitetssjukhuset, 17176 Stockholm, 23 Sweden 24 4. Department of Laboratory Medicine, Division of Pathology, Section of Perinatal 25 Pathology, Karolinska Institutet, 14186 Stockholm, Sweden 26 5. Obstetrics and Maternal Foetal Medicine Institute for Women's Health, University 27 College, WC1E 6HX London, United Kingdom 28 6. Finnish Red Cross Blood Service, 00310 Helsinki, Finland 29 30 7. Clinical Immunology and Transfusion Medicine, Karolinska University Hospital, 31 14186 Stockholm, Sweden 32 8. Dept of Laboratory Medicine, Karolinska Institutet, 17176 Stockholm, Sweden 33 9. Department of Laboratory Medicine, University Hospital North Norway, 9038

34 Tromsø, Norway http://bmjopen.bmj.com/ 35 10. Institute of Medical Biology, Faculty of Health Sciences, University of Tromsø, 9038 36 Tromsø, Norway 37 11. Department of Obstetrics and Gynaecology, Division of Clinical Science, Intervention 38 and Technology, Karolinska Institutet, 17176 Stockholm, Sweden 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43

44 45 46 47 48 49 50 51 52 Word count: 3258 53 54 55 56 57 58 59 60 1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Abstract 4 5 Objective To characterize pregnancies where the fetus or neonate was diagnosed with fetal 6 7 and neonatal alloimmune thrombocytopenia (FNAIT) and suffered from intracranial 8 9 10 hemorrhage (ICH), with special focus on time of bleeding onset. 11 12 Design Observational cohort study of all recorded cases of ICH caused by FNAIT from the 13 14 international No IntraCranial Hemorrhage (NOICH) registry during the period 2001-2010. 15 For peer review only 16 Setting Thirteen tertiary referral centres from nine countries across the world. 17 18 Participants 37 mothers and 43 children of FNAIT pregnancies complicated by fetal or 19 20 21 neonatal ICH identified from the NOICH registry was included if FNAIT diagnosis and ICH 22 23 was confirmed. 24 25 Primary and secondary outcome measures Gestational age at onset of ICH, type of ICH 26 27 and clinical outcome of ICH were the primary outcome measures. General maternal and 28 29 30 neonatal characteristics of pregnancies complicated by fetal/ neonatal ICH were secondary 31 32 outcome measures. 33

34 Results From a total of 592 FNAIT cases in the registry, 43 confirmed cases of ICH due to http://bmjopen.bmj.com/ 35 36 FNAIT were included in the study. The majority of bleedings (23/43, 54%) occurred before 37 38 28 gestational weeks and often affected the first born child (27/43, 63%). One third (35%) of 39 40 41 the children died within 4 days after delivery. 23 (53%) children survived with severe 42 on October 2, 2021 by guest. Protected copyright. 43 neurological disabilities and only five (12%) were alive and well at time of discharge. 44 45 Antenatal treatment was not given in most (91%) cases of fetal/ neonatal ICH. 46 47 Conclusions ICH caused by FNAIT often occurs during 2nd trimester and the clinical 48 49 50 outcome is poor. In order to prevent ICH caused by FNAIT, at risk pregnancies must be 51 nd 52 identified and prevention and/or interventions should start early in the 2 trimester. 53 54 Word count: 267 55 56 57 58 59 60 2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Article summary 4 5 Article focus 6 7 • Fetal and neonatal alloimmune thrombocytopenia (FNAIT) is the most common cause 8 9 10 of intracranial hemorrhage (ICH) in relation to thrombocytopenia in term born infants. 11 12 Gestational age at time of bleeding onset is important when antenatal treatment 13 14 options are discussed. No study has specifically addressed this before. 15 For peer review only 16 • From an international registry of almost 600 FNAIT cases, all recorded cases of ICH 17 18 19 were studied in detail. The aim of the study was to characterize pregnancies where the 20 21 fetus or neonate suffered from ICH caused by FNAIT, with special focus on 22 23 gestational age at time of bleeding onset. 24 25 Key messages 26 27 28 • The majority of ICH bleedings occurred by the end of the second trimester and clinical 29 30 outcome was devastating for most cases. Possible interventions to reduce risk of ICH 31 32 therefore need to be introduced before the 20th week of gestation. 33

34 • The majority ICH cases occurred in the first-born child. Most ICH cases will therefore http://bmjopen.bmj.com/ 35 36 37 not be recognized in time for treatment if we do not identify pregnancies at risk before 38 39 the first child is born. 40 41 • IVIG treatment during the subsequent pregnancy seemed to be protective with regard 42 on October 2, 2021 by guest. Protected copyright. 43 to ICH in most cases, reducing the ICH recurrence risk from 79% as previously 44 45 reported, to 11%. 46 47 48 Strengths and limitations 49 50 This is the largest study to date on ICH due to FNAIT. For the first time, gestation age at 51 52 bleeding onset was assessed in depth. Limitations of this study include that it is a 53 54 retrospective study, being subject to confounding and information bias. 55 56 57 58 59 60 3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Introduction 4 5 Fetal and neonatal alloimmune thrombocytopenia (FNAIT) is the most common cause of 6 1 7 intracranial haemorrhage (ICH) in relation to thrombocytopenia in term born infants. FNAIT 8 9 is caused by maternal alloantibodies directed against fetal platelets due to incompatibility in 10 11 human platelet antigens (HPAs). ICH due to FNAIT is reported to occur in 1: 12,500 – 25,000 12 13 births.2 The clinical outcome is often more severe than neonatal ICH from other causes.1;3 The 14 15 For peer review only4 16 ICH recurrence rate in subsequent pregnancies is reported to be 79%. Therapies have been 17 18 developed that can reduce the incidence of ICH, such as weekly high dose intravenous 19 20 immunoglobulin (IVIG).5;6;7 It is therefore considered important to identify ICH caused by 21 22 FNAIT in order to treat the mother in subsequent pregnancies. Gestational age at time of 23 24 25 bleeding onset is a key factor when antenatal treatment options are discussed since treatment 26 27 to prevent ICH needs to be commenced before ICH is likely to occur. No study has 28 29 specifically addressed this. Identifying common denominators of human HPA-alloimmunized 30 31 pregnancies complicated by ICH may serve as a future tool to help identify pregnancies at 32 33 high risk.

34 http://bmjopen.bmj.com/ 35 36 The No IntraCranial Haemorrhage (NOICH) registry is a multinational registry including 592 37 38 pregnancies complicated by FNAIT from 13 tertiary referral centres across the world. This 39 40 study presents the results of an in-depth evaluation of all recorded cases of ICH caused by 41 42 FNAIT from this registry. The aim of the study was to characterize pregnancies where the on October 2, 2021 by guest. Protected copyright. 43 44 45 fetus or neonate suffered from ICH with special focus on clinical and laboratory 46 47 characteristics and time of bleeding onset. 48 49 50 51 Materials and methods 52 53 Study design and inclusion criteria 54 55 56 All centres that had registered FNAIT cases in the NOICH registry from 2001-2010 were 57 58 invited to participate in this observational cohort study. Pregnancies recorded in the NOICH 59 60 4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 registry as complicated by fetal or neonatal ICH were identified, and included if both the 4 5 diagnosis of FNAIT and ICH were confirmed. STROBE guidelines were followed as 6 7 appropriate. 8 9 10 11 12 Case definition 13 14 A case was defined as FNAIT and included if: 1) Incompatibility between maternal and 15 For peer review only 16 paternal/ fetal HPA type was confirmed and maternal anti-HPA antibodies were detected, 2) 17 18 HPA-incompatibility between the mother and father was confirmed and the fetus/neonate 19 20 21 suffered ICH and 3) anti-HPA antibodies were detected in the mother but data on fetal/ 22 23 paternal HPA genotype was missing. 24 25 26 27 Neuroradiologic evaluation 28 29 30 Neuroradiological images were recovered and reviewed as electronic copies in a picture 31 32 archiving and communication system (PACS) using standard display programs. One 33

34 Norwegian subject and two of the Dutch subjects had MR-studies performed in utero. All http://bmjopen.bmj.com/ 35 36 available neuroradiological images were re-evaluated for this study by an experienced 37 38 independent paediatric neuroradiologist (OF). 39 40 41 The focus of the neuroradiological evaluation was primarily to confirm the ICH diagnosis, 42 on October 2, 2021 by guest. Protected copyright. 43 secondly to use imaging to assist in assessing time of bleeding onset and finally to classify the 44 45 type of bleeding. When images were not available, written reports of the imaging evaluations 46 47 for the patient files by others were used to evaluate if the ICH diagnosis was correct and the 48 49 50 subject could be included in the study. 51 52 In cases where information from normal imaging preceding abnormal imaging results was 53 54 available we were certain of a window in time during which the hemorrhage did occur. In 55 56 most subjects we did not have the benefit of initial normal imaging. Here we used instead well 57 58 59 60 5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 8 3 recognized imaging principles in judging the age of hemorrhage from its appearance on CT 4 5 or MR-imaging and estimated time of onset of bleed from this assessment. According to the 6 7 classification commonly used in the study of causes of cerebral palsy we classified the 8 9 10 hemorrhages as either intraventricular (IVH) and/or periventricular (PVH) or as parenchymal, 11 9 12 in brain parenchyma not associated with the central ventricular system. The 13 14 neuroradiological information of likely time of onset was matched to all other clinical 15 For peer review only 16 information available. 17 18 In cases where the fetus or neonate died, autopsy reports were retrieved and studied to 19 20 21 evaluate whether ICH diagnosis was certain or unlikely. All reports from post-mortem 22 23 examinations were evaluated by an experienced perinatal pathologist (NP). Only cases 24 25 evaluated to be certain ICH were included. 26 27 28 29 30 Laboratory data 31 32 All laboratory data except maternal anti-HPA-1a antibody levels were collected from the 33

34 NOICH registry database (www.NOICH.org). http://bmjopen.bmj.com/ 35 36 37 Maternal anti-HPA-1a antibody levels, except for the Finnish cases (four pregnancies), were 38 39 measured at the National reference laboratory of clinical platelet immunology in Tromsø, 40 41 10 42 Norway by quantitative MAIPA. Reproducibility between Norwegian and Finnish on October 2, 2021 by guest. Protected copyright. 43 44 quantitation was secured by double analysis of some sera samples in both Norway and 45 46 Finland. In cases where several anti-HPA-1a antibody level measurements were available, the 47 48 highest maternal anti-HPA-1a antibody level measured during pregnancy or postpartum was 49 50 51 included in the study. 52 53 54 The lowest platelet count recorded in the fetus or newborn before any platelet transfusions 55 56 were given, was included. 57 58 59 60 6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Clinical data 4 5 Clinical data was mainly collected from the NOICH registry database (www.NOICH.org). 6 7 Additional clinical information was retrieved from the original medical records by each 8 9 10 country coordinator. 11 12 The pregnancy where ICH was detected for the first time was referred to as the index case. A 13 14 subsequent ICH event was defined if ICH was recorded in any subsequent pregnancies after 15 For peer review only 16 the index ICH case. 17 18 19 20 21 Statistics 22 23 All data were analysed using SPSS software (Version 18.0 SPSS Inc., Chicago, IL). P<0.05 24 25 was considered significant. Means with 95% confidence intervals (CI) or median with range 26 27 were calculated for all continuous variables. Independent sample t-test and Kruskal-Wallis 28 29 30 test were used as appropriate. Oneway ANOVA test was used to test clinical outcome in 31 32 relation to time of bleeding onset. 33

34 http://bmjopen.bmj.com/ 35 36 Results 37 38 From the NOICH registry recording 592 FNAIT cases, 66 pregnancies from 57 women were 39 40 41 initially identified with suspected ICH. In 23 cases ICH or FNAIT diagnosis could not be 42 on October 2, 2021 by guest. Protected copyright. 43 confirmed. In total, 43 confirmed cases of ICH due to FNAIT were recorded from 37 mothers. 44 45 The 43 cases studied were included from The Netherlands, Finland, Sweden, Norway and 46 47 United Kingdom. 48 49 50 51 52 Laboratory characteristics of ICH cases 53 54 HPA-1a alloimmunization was found to be the cause in 39 of 43 ICH cases (91%). Anti-HPA- 55 56 5b antibodies were found in two further cases. In one woman with records of two ICH cases, 57 58 59 60 7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 incompatibility for HPA-5a and -5b was confirmed and in one of her ICH pregnancies anti- 4 5 HPA-5a and-5b antibodies were detected. 6 7 Data on maternal anti-HPA-1a antibody levels during pregnancy or after delivery were 8 9 10 available for 15 (35%) of the FNAIT pregnancies complicated by ICH. The median (range) 11 12 highest anti-HPA-1a antibody level was 105 IU/mL (39 to 264 IU/mL). 13 14 The median (range) lowest fetal/ neonatal platelet count for ICH cases was 8x109/L (1 to 27). 15 For peer review only 16 In comparison, median (range) neonatal platelet count for previous or subsequent pregnancies 17 18 where no ICH was detected, was 17x109/L (1 to 199) and significantly higher when compared 19 20 21 with ICH cases (Kruskal-Wallis test, P=0.004). 22 23 24 25 Clinical characteristics of ICH pregnancies 26 27 Maternal characteristics for all ICH cases are shown in Table 1. In the group of the 37 index 28 29 30 cases, 26 (70%) cases were first born children. However, the mothers were primigravidae in 31 nd 32 only 10/37 (27%) index cases. Ten mothers (23%) experienced one or more 2 trimester 33

34 miscarriages (altogether 20). http://bmjopen.bmj.com/ 35 36 Antenatal treatment was given in 4/43 (9%) of the pregnancies complicated by fetal/ neonatal 37 38 ICH. In 3 cases IVIG was given, two of these fetuses received additional intrauterine platelet 39 40 41 transfusions. One mother received corticosteroids as single treatment from 6 weeks gestation 42 on October 2, 2021 by guest. Protected copyright. 43 onwards, due to two previous 2nd trimester miscarriages (not identified as FNAIT related). For 44 45 the 37 index cases, only one mother received antenatal treatment. In this case, IVIG treatment 46 47 was started from week 7 due to autoimmune thrombocytopenia in the mother. 48 49 50 Twenty out of 43 mothers with pregnancies complicated by fetal/ neonatal ICH were 51 52 delivered by caesarean section (CS); CS was performed in nine cases with abnormal fetal 53 54 cerebral ultrasound as main indication. 55 56 57 58 59 60 8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 A high proportion of the pregnancies complicated by ICH were ended preterm: 29 out of 43 4 5 children (67%) were born before 37 weeks and 4 children were born before 28 weeks. 6 7 Abnormal US scan with detection of ICH was the main reason for premature delivery (Table 8 9 10 2). 11 12 Five (12%) of the infants died in utero and one (2%) died during labour. Neonatal 13 14 characteristics of the live-born children are shown in Table 3. Nine children died within the 15 For peer review only 16 first four days after delivery. Most of the survivors developed neurological sequelae: Eight 17 18 infants were diagnosed with cerebral palsy, ten were reported to be moderate/ severely 19 20 21 mentally retarded or severely disabled, seven neonates were reported to have epilepsy and 22 23 four were blind or with severely reduced vision. In addition, one case of autism and one case 24 25 of impaired hearing were reported. More than one neurological complication was reported in 26 27 several cases. Five (14%) of surviving neonates with ICH were reported to be alive and well 28 29 30 at time of discharge after delivery, but data on long-term clinical outcome is missing. 31 32 The fetuses/ neonates were male in the majority (65%) of ICH cases. There was no significant 33

34 difference in maternal anti-HPA-1a antibody levels, birth weight, APGAR scores or platelet http://bmjopen.bmj.com/ 35 36 counts when comparing boys and girls (Independent sample T-test, P>0.05). 37 38 Ten (23%) ICH neonates were below the 10th percentile for birth weight and defined as small 39 40 41 for gestational age (SGA) according to standard growth curves. All SGA cases except one 42 on October 2, 2021 by guest. Protected copyright. 43 were boys. 44 45 46 47 Bleeding characteristics 48 49 50 Neuroradiological studies were used to review and confirm 19 ICH cases. The remaining 24 51 52 ICH cases were confirmed using descriptions of radiographic material by local radiologists. 53 54 The estimated time of bleeding onset for each ICH case is shown in Figure 1a. The figure 55 56 demonstrates that many bleedings started early, and that in most cases several weeks passed 57 58 59 60 9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 between bleeding onset and delivery. More than half (23/43, 54%) of the bleedings happened 4 5 around gestational week 28 or earlier (Figure 1b). In 16 (70%) of the 23 cases with bleeding 6 7 onset before the 3rd trimester we do not know the exact time of bleeding onset, only that the 8 9 10 bleeding started no later than 28 weeks (Figure 1a). Twenty-nine out of 43 bleedings (67%) 11 12 started before 34 gestational weeks (Figure 1b). No cases of intrapartum ICH bleedings were 13 14 confirmed. The time of bleeding onset in the 10 primigravid cases were not different from the 15 For peer review only 16 multigravid ICH cases. 17 18 There was no difference in clinical outcome in relation to time of bleeding onset (Figure 2, 19 20 21 Oneway ANOVA, P=0.75). However, it is noteworthy that perinatal death occurred in two 22 23 girls (14%) where the bleedings were found to have happened before 30 weeks. In boys, 24 25 perinatal death occurred in 13 (46%) cases with bleedings happening also at a later gestational 26 27 age (data not shown). 28 29 30 Intraparenchymal hemorrhages were found in 11 out of 13 cases where the bleeding occurred 31 32 during the third trimester or after delivery. See Figure 2a for illustration of intraparenchymal 33

34 hemorrhage. http://bmjopen.bmj.com/ 35 36 In five cases, multiple bleeding episodes were found. All second bleedings occurred after 33 37 38 weeks (range 33 to 37 weeks). These five cases were all due to HPA-1a alloimmunization. 39 40 41 Fifteen ICH cases were classified as intraparenchymal and 13 cases as IVH/ PVH (Figure 2b). 42 on October 2, 2021 by guest. Protected copyright. 43 Five cases were classified as miscellaneous (Figure 2c), whereas 10 cases could not be 44 45 classified. Among cases where ICH was found to occur before 28 weeks, all but one case was 46 47 found to be IVH/PVH. 48 49 50 51 52 Treatment and clinical outcome in subsequent pregnancies 53 54 55 56 57 58 59 60 10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 ICH was detected in six (23%) subsequent pregnancies. One woman had records of ICH in 4 5 two subsequent pregnancies after the index ICH case. In most subsequent pregnancies (20/26, 6 7 77%), no ICH was found. 8 9 10 In 19/ 26 (73%) of subsequent pregnancies, the mothers received antenatal treatment (IVIG). 11 12 In the six cases where ICH was detected in subsequent pregnancies, three received IVIG 13 14 treatment. However, in one of these cases IVIG treatment was started after ICH was detected. 15 For peer review only 16 Therefore, IVIG treatment failed to prevent ICH in two (11%) out of 19 cases. These two 17 18 treatment failures were from the same woman. It should be commented that the obstetrical 19 20 21 history of this woman is particularly severe with regards to FNAIT complications. The ICH 22 23 recurrence rate was therefore 11% in IVIG treated pregnancies. Compared with historical data 24 25 reporting 79% risk of ICH recurrence in FNAIT,4 our data indicate that IVIG was effective in 26 27 preventing ICH. 28 29 30 In seven cases, no IVIG treatment was given during a subsequent pregnancy after ICH was 31 32 detected in a previous pregnancy. Five of these untreated pregnancies come from Norway, 33

34 where IVIG is not routinely given as treatment during HPA alloimmunized pregnancies. ICH http://bmjopen.bmj.com/ 35 36 occurred in three of these seven cases. 37 38

39 40 41 Discussion 42 on October 2, 2021 by guest. Protected copyright. 43 The main findings of this study are that the majority of ICH bleedings occurred by the end of 44 45 the second trimester and that clinical outcome was devastating for most cases. The high 46 47 frequency of bleedings occurring before 28 weeks indicates that the fetus may be severely 48 49 50 affected already in the second trimester. 51 52 This is the largest study to date on fetal/ neonatal ICH caused by FNAIT. For the first time, 53 54 time of bleeding onset was assessed using clinical information together with radiographic 55 56 imaging and autopsy reports. The in-depth study of both laboratory and clinical information 57 58 59 60 11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 was done in close collaboration between obstetricians, immunologists, perinatal pathologists 4 5 and specialists in neuroradiology. Limitations of this study include that it is a retrospective 6 7 cohort study, being subject to confounding and information bias. A bias toward inclusion of 8 9 10 the more severe ICH phenotype may be considered. The ICH cases in this study were 11 12 collected from five different countries and therefore reflects several institution`s clinical 13 14 experiences. There is obvious heterogeneity in antenatal treatment for FNAIT between these 15 For peer review only 16 countries. However, since most of the patients affected by foetal ICH did not receive antenatal 17 18 treatment, we consider the study population uniform and representative of a larger population. 19 20 21 Some earlier studies have suggested the onset of bleedings to be in the third rather than the 22 11;12 23 second trimester, and are in variance with this study. The judgment of the onset of 24 25 cerebral bleeding in this study was cautious, and the onset was probably in many cases even 26 27 earlier than we report. However, previous studies included too few cases to address this 28 29 30 question in an adequate manner. Also, and more importantly, other studies report the 31 32 gestational age when the ICH was diagnosed, and did not assess when the bleeding may have 33

34 occurred. Most of the ICH cases occurred in boys, and the bleedings were more often lethal http://bmjopen.bmj.com/ 35 36 when the fetus was male. The finding that 80% of the SGA neonates in our population were 37 38 boys supports the recently published observation that birth weight in boys, but not in girls, 39 40 13 41 was associated with maternal anti-HPA-1a antibodies. Data on maternal anti-HPA-1a 42 on October 2, 2021 by guest. Protected copyright. 43 antibody levels in FNAIT pregnancies complicated by fetal/ neonatal ICH have not been 44 45 published before. 46 47 Fetal ICH due to FNAIT often occurred in the first child. Most ICH cases will therefore not 48 49 50 be recognized in time for treatment or prophylactic measures if we do not identify 51 52 pregnancies at risk before the onset of the bleeding. This finding therefore challenges the 53 54 current management strategy where antenatal treatment is given in subsequent pregnancies 55 56 after FNAIT has been diagnosed in the first child. A majority of these children will suffer 57 58 59 60 12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 from bleeding already in the second trimester or in the early part of the third trimester. 4 5 Possible interventions to reduce risk of ICH need to be introduced before the 20th week of 6 7 gestation. In most children the ICH was either fatal or induced severe disabilities. Our 8 9 10 findings therefore support the idea of identifying HPA-1bb mothers with anti-HPA-1a 11 12;14 12 antibodies in all pregnancies and working towards a prophylactic approach to prevent the 13 14 immune response against HPA-1a.15;16 The rate of prematurity and CS was high in our study. 15 For peer review only 16 Many of these patients required delivery before term due to problems related to the fetal ICH 17 18 such as fetal distress, lack of fetal movements, abnormal ultrasound scan etc. Thus, the 19 20 21 present study does not indicate that FNAIT per se is associated with an increased risk for 22 23 prematurity. 24 25 There were no confirmed cases of ICH occurring intrapartum in this study, and only two 26 27 bleedings occurred after delivery. This could suggest that mode of delivery may not be so 28 29 30 important in the prevention of ICH. However, the high CS rate in this study population may 31 32 have contributed to the low number of intrapartum bleedings. Whether or not delivery by CS 33 12;17

34 prevents ICH needs to be further addressed. The present study suggests that IVIG http://bmjopen.bmj.com/ 35 36 treatment during pregnancy is protective in regard to ICH in most cases. However, it is an 37 38 open question whether IVIG would have protected the first born child from ICH, or whether 39 40 41 there is a genuine increased risk of ICH in the first born child. Further, it remains to be 42 on October 2, 2021 by guest. Protected copyright. 43 established if there is also a milder phenotype of ICH with discrete symptoms and better 44 45 outcome. This question can only be addressed in prospective studies including general 46 47 screening and repeated fetal US examinations. The maternal anti-HPA-1a antibody levels 48 49 50 were extremely high among ICH cases in this study. Maternal anti-HPA-1a antibody levels 51 52 may therefore be useful to identify pregnancies at risk for fetal ICH, but these findings need to 53 54 be evaluated in larger prospective studies. Finally, why boys seem to be more susceptible to 55 56 maternal anti-HPA-1a antibodies is currently not known and needs further investigation. 57 58 59 60 13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 4 5 Acknowledgements 6 7 We thank Dr. Eline S. Van den Akker, MD, PhD, Amsterdam, the Netherlands for her initial 8 9 10 work setting up the NOICH database and registry. No compensation was received for this 11 12 contribution. 13 14 15 For peer review only 16 Contributors 17 18 19 20 MW, AH and HT designed the study. HT, AH and MW carried out the data analyses. HT 21 22 drafted the manuscript. All authors commented on the drafts of the manuscript and 23 24 contributed to the final version. AH and MW are guarantors of the study. All authors, external 25 26 and internal, had full access to all of the data (including statistical reports and tables) in the 27 28 study and can take responsibility for the integrity of the data and the accuracy of the data 29 30 31 analysis. 32 33

34 http://bmjopen.bmj.com/ 35 36 37 Funding 38 39 40 ALD have support from UCLH/UCL by funding from the Department of Health’s NIHR 41 42 Biomedical Research Centres funding scheme for the submitted work. The funder had no part on October 2, 2021 by guest. Protected copyright. 43 44 in the design or execution of the study or the analysis and interpretation of the results. 45 46 47 48 49 Conflicts of interest 50 51 All authors have completed the Unified Competing Interest form at 52 53 www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and 54 55 declare that (1) no support from any organisation for the submitted work; (2) AH has stocks 56 57 in Prophylix Pharma AS trying to develop a prophylaxis against HPA-1a immunization that 58 59 60 14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 might have an interest in the submitted work in the previous 3 years; (3) their spouses, 4 5 partners, or children have no financial relationships that may be relevant to the submitted 6 7 work; and (4) none of the authors have non-financial interests that may be relevant to the 8 9 10 submitted work. 11 12 13 14 Details of ethics approval 15 For peer review only 16 All ethics approvals were obtained on a National level, based on local regulations: Sweden, 17 18 “Regionala etikprövningsnämnden”, approval number 2005/2:1. Norway, the Regional 19 20 21 Committee for Medical Research Ethics, North Norway, approval number 2006/1585. UK: 22 23 The study was done as part of institutional performance, and it received institutional review 24 25 board exemption. The Netherlands: Ethics approval was waived by the IRB of the LUMC. 26 27 Finland: The study was based on the NAIT registry kept by the Finnish Red Cross Blood 28 29 30 Service according to the Finnish law. 31 32 33

34 http://bmjopen.bmj.com/ 35 Copyright 36 37 38 The corresponding author has the right to grant on behalf of all authors and does grant on 39 40 41 behalf of all authors, an exclusive licence (or non exclusive for government employees) on a 42 on October 2, 2021 by guest. Protected copyright. 43 worldwide basis to the BMJ Publishing Group Ltd to permit this article to be published in 44 45 BMJ editions and any other BMJPGL products and sublicences such use and exploit all 46 47 subsidiary rights, as set out in our licence. 48 49 50 51 Data sharing 52 53 54 No additional data available 55 56 57 58 59 60 15 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Reference List 4 5 (1) Mao C, Guo J, Chituwo BM. Intraventricular haemorrhage and its prognosis, prevention and treatment 6 in term infants. J Trop Pediatr 1999; 45(4):237-240. 7 (2) Williamson LM, Hackett G, Rennie J et al. The natural history of fetomaternal alloimmunization to the 8 platelet-specific antigen HPA-1a (PlA1, Zwa) as determined by antenatal screening. Blood 1998; 9 92:2280-2287. 10 11 (3) Jocelyn LJ, Casiro OG. Neurodevelopmental outcome of term infants with intraventricular hemorrhage. 12 Am J Dis Child 1992; 146(2):194-197. 13 14 (4) Radder CM, Brand A, Kanhai HH. Will it ever be possible to balance the risk of intracranial 15 haemorrhageFor in fetal peeror neonatal alloimmune review thrombocytopenia against only the risk of treatment strategies 16 to prevent it? Vox Sang 2003; 84(4):318-325. 17 18 (5) Kamphuis MM, Oepkes D. Fetal and neonatal alloimmune thrombocytopenia: prenatal interventions. 19 Prenat Diagn 2011; 31(7):712-719. 20 21 (6) Kanhai HH, Porcelijn L, Engelfriet CP et al. Management of alloimmune thrombocytopenia. Vox Sang 2007; 93(4):370-385. 22 23 (7) Pacheco LD, Berkowitz RL, Moise KJ et al. Fetal and neonatal alloimmune thrombocytopenia. A 24 management algorithm based on risk stratification. Obstet Gynecol 2012; 118(5):1157-1163. 25 26 (8) Osborn AG. Imaging of intracranial hemorrhage. Diagnostic neuroradiology. Mosby; 1994. 158-173. 27 28 (9) Bax M, Tydeman C, Flodmark O. Clinical and MRI correlates of cerebral palsy. JAMA 2012; 29 296(13):1602-1608. 30 31 (10) Kiefel V, Santoso S, Weisheit M et al. Monoclonal antibody--specific immobilization of platelet 32 antigens (MAIPA): a new tool for the identification of platelet-reactive antibodies. Blood 1987; 33 70:1722-1726.

34 http://bmjopen.bmj.com/ 35 (11) Spencer JA, Burrows RF. Feto-maternal alloimmune thrombocytopenia: a literature review and 36 statistical analysis. Aust N Z J Obstet Gynaecol 2001; 41:45-55. 37 (12) Kamphuis MM, Paridaans N, Porcelijn L et al. Screening in pregnancy for fetal or neonatal alloimmune 38 thrombocytopenia: systematic review. BJOG 2010; 117(11):1335-1343. 39 40 (13) Tiller H, Killie MK, Husebekk A. Platelet antibodies and fetal growth: Maternal antibodies against fetal 41 platelet antigen 1a are strongly associated with reduced birthweight in boys. Acta obstetricia et 42 gynecologica Scandinavica 2011; 91:79-86. on October 2, 2021 by guest. Protected copyright. 43 44 (14) Kjeldsen-Kragh J, Killie MK, Tomter G et al. A screening and intervention program aimed to reduce 45 mortality and serious morbidity associated with severe neonatal alloimmune thrombocytopenia. Blood 46 2007; 110(3):833-839. 47 48 (15) Kjeldsen-Kragh J, Ni H, Skogen B. Towards a prophylactic treatment of HPA-related foetal and 49 neonatal alloimmune thrombocytopenia. Curr Opin Hematol 2012; 19(6):469-474. 50 51 (16) Tiller H, Killie MK, Chen P et al. Toward a prophylaxis against fetal and neonatal alloimmune 52 thrombocytopenia: induction of antibody-mediated immune suppression and prevention of severe clinical complications in a murine model. Transfusion 2012; 52(7):1446-1457. 53 54 (17) van den AE, Oepkes D, Brand A et al. Vaginal delivery for fetuses at risk of alloimmune 55 thrombocytopenia? BJOG 2006; 113(7):781-783. 56 57 58 59 60 16 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 4 Figure legends 5 6 Figure 1 7 8 a) Estimated time period for onset of ICH is shown for 41 cases of ICH caused by FNAIT. An 9 10 arrowhead to the left indicates that the earliest time of onset cannot be estimated, only that the 11 12 13 bleeding occurred before the gestational age indicated on the x-axis. 14 15 For peer review only 16 17 b) Time of bleeding onset is divided into three groups (<28, 28-34 and >34 gestational weeks) 18 19 and the frequency in each group is shown on the y-axis. Clinical outcome is categorized as 20 21 22 dead, alive with neurological sequelae or alive and well. 23 24 25 26 Figure 2 27 28 a) CT scan shows a very large intraparenchymal hemorrhage with mass effect and occupying 29 30 most of the right frontal lobe in a term baby. Note a second but smaller hemorrhage on the left 31 32 33 side adjacent to the ventricle but separate from this. The attenuation of both hemorrhages

34 http://bmjopen.bmj.com/ 35 indicates that the hemorrhage is several weeks old at the time of imaging. We estimated the 36 37 right hemorrhage to be 4 - 6 weeks old and the left still a couple of weeks older. The 38 39 pregnancy was considered normal, and no IVIG treatment was given. A live boy was born 40 41 9 42 vaginally at 40 gestational weeks with petecchiae. The platelet count was 6x10 /L. on October 2, 2021 by guest. Protected copyright. 43 44 45 46 b) This CT shows marked ventriculomegaly, partly due to hydrocephalus, partly due to loss of 47 48 brain tissue in the left hemisphere. Note the very large ventricle on the left side occupying 49 50 51 most of the left hemicranium. The shape of the left lateral ventricle is irregular and that of 52 53 residual from a previous intraventricular hemorrhage with a paraventricular atrophic defect 54 55 caused by an old periventricular hemorrhagic infarction. This pattern is pathognomonic for 56 57 this condition even though no actual blood can be detected. The interpretation of this image is 58 59 60 17 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 that of an intraventricular hemorrhage associated with a periventricular hemorrhagic 4 5 infarction timed at 28 – 30 gestational weeks or earlier. Hydrocephalus was diagnosed 6 7 intrapartum because of breech presentation. A live boy was born at term with a platelet count 8 9 9 10 of 4x10 /L. The child has severe cerebral palsy and a hydrocephalus shunt. 11 12 13 14 c) The CT scan shows multiple focal intraparenchymal hemorrhages throughout both cerebral 15 For peer review only 16 hemispheres. Most hemorrhages are quite small. Note also the extensive extracranial 17 18 subgaleal hemorrhage overlying the right hemicranium. All bleedings are of same age and 19 20 21 maximum seven days old. It was noted that the fetus was small on US scan at 25 weeks, but 22 23 otherwise the pregnancy was considered normal. At 42 weeks a live boy was born vaginally, 24 25 with multiple petecchiaes and multiple retinal bleedings. The platelet count at delivery was 26 27 14x109/L, with nadir value 8x109/L. He received platelet transfusions. At time of discharge 28 29 30 the boy was described as alive and well, but we do not have data on long-term clinical 31 32 outcome. 33

34 http://bmjopen.bmj.com/ 35 36 37 38

39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 18 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Table 1 Maternal characteristics of ICH-affected pregnancies 4 5 6 7 Maternal characteristic 8 9 10 Maternal age in years, mean (95% CI) 29.3 (27.6-31.0) 11 12 Obstetrical history 13 14 Primigravida, n (%) 10 (23) 15 For peer review only 16 First-born child, n (%) 27 (63) 17 18 nd 19 2 trimester miscarriage, n (%) 10 (23) 20 21 Pre-eclampsia in this pregnancy, n (%) 3 (7) 22 23 Vaginal delivery of ICH neonate, n (%) 22* (51) 24 25 Caesarean section of ICH neonate, n (%) 20* (47) 26 27 28 Fetal/ maternal treatment 29 30 IVIG, n (%) 4 (9) 31 32 Steroids, n (%) 1 (2) 33

34 http://bmjopen.bmj.com/ Intrauterine platelet transfusion, n (%) 3 (7) 35 36 37 No treatment 36 (84) 38 39 *Mode of delivery was not known for one ICH pregnancy 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 19 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Table 2 Preterm birth 4 5 Reason for preterm delivery Number of pregnancies (% of total) 6 7 Abnormal US scan (ICH detected) 15* (52) 8 9 10 Intrauterine fetal demise 5* (17) 11 12 Intrauterine growth restriction/ pre-eclampsia 3 (10) 13 14 Spontaneous preterm birth 4 (14) 15 For peer review only 16 Not known 2 (7) 17 18 19 Total 29** (100) 20 21 *One case was a complication during cordocentesis 22 **Including one twin pregnancy 23 24 25 26

27 28 29 Table 3 Clinical outcome characteristics of ICH cases 30 31 Characteristic Result 32 33 All ICH cases

34 http://bmjopen.bmj.com/ 35 Gestational age at delivery, median (range) in weeks 35 (23-42) 36 37 38 Birth weight, mean (SD) in grams 2274 (832) 39 40 Sex, female/ male 14/28 41 42 Stillborn 6 (14) on October 2, 2021 by guest. Protected copyright. 43 44 45 Live-born children 37 (86) 46 47 APGAR score <7 after 5 minutes, n (%) 8 (23) 48 49 Lowest platelet count (range) x109/L* 8.9 (1-27) 50 51 Neonatal death, n (%) 9 (21) 52 53 54 Alive and well at discharge, n (%) 5 (12) 55 56 Alive with neurological sequelae, n (%) 23 (53) 57 58 *Data available for 32 pregnancies 59 60 20 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 STROBE Statement—Checklist of items that should be included in reports of cohort studies 2 Item 3 No Recommendation 4 Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the abstract 5 6 Ok 7 (b) Provide in the abstract an informative and balanced summary of what was done 8 and what was found 9 Ok 10 11 Introduction 12 Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 13 Ok 14 15 Objectives For3 Statepeer specific objectives, review including any prespecified only hypotheses 16 Ok 17 Methods 18 Study design 4 Present key elements of study design early in the paper 19 20 Ok 21 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, 22 exposure, follow-up, and data collection 23 Ok 24 a 25 Participants 6 ( ) Give the eligibility criteria, and the sources and methods of selection of 26 participants. Describe methods of follow-up Ok 27 28 (b) For matched studies, give matching criteria and number of exposed and unexposed 29 NA 30 31 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and effect 32 modifiers. Give diagnostic criteria, if applicable Ok 33 Data sources/ 8* For each variable of interest, give sources of data and details of methods of

34 http://bmjopen.bmj.com/ measurement assessment (measurement). Describe comparability of assessment methods if there is 35 36 more than one group Ok 37 Bias 9 Describe any efforts to address potential sources of bias 38 Potential sources of bias are included in the discussion 39 Study size 10 Explain how the study size was arrived at 40 All recorded cases of ICH caused by FNAIT from the international No 41 IntraCranial Haemorrhage (NOICH) registry during the period 2001-2010, were 42 included. From the NOICH registry recording 592 FNAIT cases, 66 pregnancies on October 2, 2021 by guest. Protected copyright. 43 from 57 women were initially identified with suspected ICH. In 23 cases ICH or 44 FNAIT diagnosis could not be confirmed. In total, 43 confirmed cases of ICH due 45 to FNAIT were recorded. 46 Quantitative 11 Explain how quantitative variables were handled in the analyses. 47 variables Means with 95% confidence intervals (CI) or median with range were calculated 48 for all continuous variables 49 If applicable, describe which groupings were chosen and why NA 50 Statistical methods 12 (a) Describe all statistical methods, including those used to control for confounding 51 Ok 52 (b) Describe any methods used to examine subgroups and interactions NA 53 c 54 ( ) Explain how missing data were addressed 55 If necessary data to confirm the FNAIT or ICH diagnosis was missing, the case 56 was excluded from the final study population 57 (d) If applicable, explain how loss to follow-up was addressed NA 58 (e) Describe any sensitivity analyses NA 59 Results 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml1 BMJ Open Page 22 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially 2 eligible, examined for eligibility, confirmed eligible, included in the study, completing 3 follow-up, and analysed Ok 4 5 (b) Give reasons for non-participation at each stage OK 6 (c) Consider use of a flow diagram Not considered to be necessary 7 Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and 8 information on exposures and potential confounders Ok 9 10 (b) Indicate number of participants with missing data for each variable of interest 11 Was not included in manuscript due to space considerations, but may be 12 provided upon request 13 (c) Summarise follow-up time (eg, average and total amount) NA 14 Outcome data 15* Report numbers of outcome events or summary measures over time NA 15 For peer review only 16 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and 17 their precision (eg, 95% confidence interval). Make clear which confounders were 18 adjusted for and why they were included 19 Ok. Confounder-adjusted estimates were n.a. 20 b 21 ( ) Report category boundaries when continuous variables were categorized NA 22 (c) If relevant, consider translating estimates of relative risk into absolute risk for a 23 meaningful time period NA 24 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity 25 26 analyses NA 27 Discussion 28 Key results 18 Summarise key results with reference to study objectives Ok 29 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or 30 31 imprecision. Discuss both direction and magnitude of any potential bias Ok 32 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, 33 multiplicity of analyses, results from similar studies, and other relevant evidence Ok

Page41 23 of 25 BMJ Open 40 39 1 38 2 37 36 3 35For peer review only 4 34 5 33 6 32 7 31 30 8 29 9 28 10 27

11 26 http://bmjopen.bmj.com/ 12 25 13 24 23 14 22 15 21 16 Patient 20 17 19 Gender unknown 18 18 17 19 16 on October 2, 2021 by guest. Protected copyright. 20 15 21 14 22 13 23 12 11 24 10 25 9 26 8 27 7 28 6 5 29 4 30 3 31 2 32 1 33 16 18 20 22 24 26 2830 32 34 36 38 40 42 34 Gestational age (weeks) 35 36 Onset of Time between bleeding Time of delivery 37 bleeding and delivery 38 For peer Malereview only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 39 40 Female 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 24 of 25 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Figure 1b 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Fig 2 4 5 a) Intraparenchymal hemorrhage 6 7 Dx 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 b) Intraventricular hemorrhage 24 25 26 Dx 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 c) Miscellaneous (Multiple intraparenchymal hemorrhages) 42 on October 2, 2021 by guest. Protected copyright. 43 44 Dx 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from

Fetal intracranial haemorrhages caused by fetal and neonatal alloimmune thrombocytopenia: An observational cohort study of 43 cases from an international multicentre registry

For peer review only Journal: BMJ Open

Manuscript ID: bmjopen-2012-002490.R1

Article Type: Research

Date Submitted by the Author: 16-Jan-2013

Primary Subject Obstetrics and gynaecology

Heading: on October 2, 2021 by guest. Protected copyright.

Secondary Subject Heading: Immunology (including allergy)

Fetal medicine < OBSTETRICS, Prenatal diagnosis < OBSTETRICS, Keywords: Ultrasonography < OBSTETRICS

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Fetal intracranial haemorrhages caused by fetal and neonatal alloimmune 4 5 thrombocytopenia: An observational cohort study of 43 cases from an international 6 multicentre registry 7 1 2 3 8 Heidi Tiller HT, Marije M. Kamphuis MMK, Olof Flodmark OF, Nikos Papadogiannakis 9 4 5 6 6 6 10 NP, Anna L David ALD, Susanna Sainio SS, Sinikka Koskinen SK, Kaija Javela KJ, 11 Agneta Wikman AW,7 Riitta Kekomaki RK,6 Humphrey H.H. Kanhai HHHK,2 Dick Oepkes 12 2 10 11 13 DO, Anne Husebekk AH, * and Magnus Westgren MW * 14 *Corresponding authors 15 For peer review only 16 17 1. Department of Obstetrics and Gynaecology, University Hospital North Norway, 9038 18 Tromsø, Norway 19 20 2. Department of Obstetrics, Leiden University Medical Center, 2253LZ Leiden, The 21 Netherlands 22 3. Neuroradiologiska kliniken, Karolinska Universitetssjukhuset, 17176 Stockholm, 23 Sweden 24 4. Department of Laboratory Medicine, Division of Pathology, Section of Perinatal 25 Pathology, Karolinska Institutet, 14186 Stockholm, Sweden 26 5. Obstetrics and Maternal Foetal Medicine Institute for Women's Health, University 27 College, WC1E 6HX London, United Kingdom 28 6. Finnish Red Cross Blood Service, 00310 Helsinki, Finland 29 30 7. Clinical Immunology and Transfusion Medicine, Karolinska University Hospital, 31 14186 Stockholm, Sweden 32 8. Dept of Laboratory Medicine, Karolinska Institutet, 17176 Stockholm, Sweden 33 9. Department of Laboratory Medicine, University Hospital North Norway, 9038

44 45 46 47 48 49 50 51 52 Word count: 3609 53 54 55 56 57 58 59 60 1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Abstract 4 5 Objective To characterize pregnancies where the fetus or neonate was diagnosed with fetal 6 7 and neonatal alloimmune thrombocytopenia (FNAIT) and suffered from intracranial 8 9 10 haemorrhage (ICH), with special focus on time of bleeding onset. 11 12 Design Observational cohort study of all recorded cases of ICH caused by FNAIT from the 13 14 international No IntraCranial Haemorrhage (NOICH) registry during the period 2001-2010. 15 For peer review only 16 Setting Thirteen tertiary referral centres from nine countries across the world. 17 18 Participants 37 mothers and 43 children of FNAIT pregnancies complicated by fetal or 19 20 21 neonatal ICH identified from the NOICH registry was included if FNAIT diagnosis and ICH 22 23 was confirmed. 24 25 Primary and secondary outcome measures Gestational age at onset of ICH, type of ICH 26 27 and clinical outcome of ICH were the primary outcome measures. General maternal and 28 29 30 neonatal characteristics of pregnancies complicated by fetal/ neonatal ICH were secondary 31 32 outcome measures. 33

34 Results From a total of 592 FNAIT cases in the registry, 43 confirmed cases of ICH due to http://bmjopen.bmj.com/ 35 36 FNAIT were included in the study. The majority of bleedings (23/43, 54%) occurred before 37 38 28 gestational weeks and often affected the first born child (27/43, 63%). One third (35%) of 39 40 41 the children died within 4 days after delivery. 23 (53%) children survived with severe 42 on October 2, 2021 by guest. Protected copyright. 43 neurological disabilities and only five (12%) were alive and well at time of discharge. 44 45 Antenatal treatment was not given in most (91%) cases of fetal/ neonatal ICH. 46 47 Conclusions ICH caused by FNAIT often occurs during 2nd trimester and the clinical 48 49 50 outcome is poor. In order to prevent ICH caused by FNAIT, at risk pregnancies must be 51 nd 52 identified and prevention and/or interventions should start early in the 2 trimester. 53 54 Word count: 267 55 56 57 58 59 60 2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Article summary 4 5 Article focus 6 7 • Fetal and neonatal alloimmune thrombocytopenia (FNAIT) is the most common cause 8 9 10 of intracranial haemorrhage (ICH) in relation to thrombocytopenia in term born 11 12 infants. Gestational age at time of bleeding onset is important when antenatal 13 14 treatment options are discussed. No study has specifically addressed this before. 15 For peer review only 16 • From an international registry of almost 600 FNAIT cases, all recorded cases of ICH 17 18 19 were studied in detail. The aim of the study was to characterize pregnancies where the 20 21 fetus or neonate suffered from ICH caused by FNAIT, with special focus on 22 23 gestational age at time of bleeding onset. 24 25 Key messages 26 27 28 • The majority of ICH bleedings occurred by the end of the second trimester and clinical 29 30 outcome was devastating for most cases. Possible interventions to reduce risk of ICH 31 32 therefore need to be introduced before the 20th week of gestation. 33

34 • The majority ICH cases occurred in the first-born child. Most ICH cases will therefore http://bmjopen.bmj.com/ 35 36 37 not be recognized in time for treatment if we do not identify pregnancies at risk before 38 39 the first child is born. 40 41 • IVIG treatment during the subsequent pregnancy seemed to be protective with regard 42 on October 2, 2021 by guest. Protected copyright. 43 to ICH in most cases, reducing the ICH recurrence risk from 79% as previously 44 45 reported, to 11%. 46 47 48 Strengths and limitations 49 50 This is the largest study to date on ICH due to FNAIT. For the first time, gestational age at 51 52 bleeding onset was assessed in depth. Limitations of this study include that it is a 53 54 retrospective study, being subject to confounding and information bias. 55 56 57 58 59 60 3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Introduction 4 5 Fetal and neonatal alloimmune thrombocytopenia (FNAIT) is the most common cause of 6 1 7 intracranial haemorrhage (ICH) in relation to thrombocytopenia in term born infants. FNAIT 8 9 is caused by maternal alloantibodies directed against fetal platelets due to incompatibility in 10 11 human platelet antigens (HPAs). ICH due to FNAIT is reported to occur in 1: 12,500 – 25,000 12 13 births.2,3 The clinical outcome is often more severe than for neonatal ICH from other 14 15 1,4 For peer review only 5 16 causes. The ICH recurrence rate in subsequent pregnancies is reported to be 79%. 17 18 Therapies have been developed that can reduce the incidence of ICH, such as weekly high 19 20 dose intravenous immunoglobulin (IVIG).6,7,8 It is therefore considered important to identify 21 22 ICH caused by FNAIT in order to treat the mother in subsequent pregnancies. Gestational age 23 24 25 at time of bleeding onset is a key factor when antenatal treatment options are discussed since 26 27 treatment to prevent ICH needs to be commenced before ICH is likely to occur. No study has 28 29 specifically addressed this. Identifying common denominators of human HPA-alloimmunized 30 31 pregnancies complicated by ICH may serve as a future tool to help identify pregnancies at 32 33 high risk.

34 http://bmjopen.bmj.com/ 35 36 The No IntraCranial Haemorrhage (NOICH) registry is a multinational registry including 592 37 38 pregnancies complicated by FNAIT from 13 tertiary referral centres across the world. This 39 40 study presents the results of an in-depth evaluation of all recorded cases of ICH caused by 41 42 FNAIT from this registry. The aim of the study was to characterize pregnancies where the on October 2, 2021 by guest. Protected copyright. 43 44 45 fetus or neonate suffered from ICH with special focus on clinical and laboratory 46 47 characteristics and time of bleeding onset. 48 49 50 51 Materials and methods 52 53 Study design and inclusion criteria 54 55 56 All centres that had registered FNAIT cases in the NOICH registry from 2001-2010 were 57 58 invited to participate in this observational cohort study. Pregnancies recorded in the NOICH 59 60 4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 registry as complicated by fetal or neonatal ICH were identified, and included if both the 4 5 diagnosis of FNAIT and ICH were confirmed. STROBE guidelines were followed as 6 7 appropriate. 8 9 10 A case was defined as FNAIT if: 1) Incompatibility between maternal and paternal/ fetal HPA 11 12 type was confirmed and maternal anti-HPA antibodies were detected, 2) HPA-incompatibility 13 14 between the mother and father was confirmed and the fetus/neonate suffered ICH and 3) anti- 15 For peer review only 16 HPA antibodies were detected in the mother but data on fetal/ paternal HPA genotype was 17 18 missing. 19 20 21 Neuroradiological images were recovered and reviewed as electronic copies in a picture 22 23 archiving and communication system (PACS) using standard display programs. One 24 25 Norwegian subject and two of the Dutch subjects had MR-studies performed in utero. All 26 27 available neuroradiological images were re-evaluated for this study by an experienced 28 29 30 independent paediatric neuroradiologist (OF). The focus of the neuroradiological evaluation 31 32 was primarily to confirm the ICH diagnosis, secondly to use imaging to assist in assessing 33

34 time of bleeding onset and finally to classify the type of bleeding. When images were not http://bmjopen.bmj.com/ 35 36 available, written reports of the imaging evaluations for the patient files by others were used 37 38 to evaluate if the ICH diagnosis was correct. Cases where an ICH could not be confirmed 39 40 41 were not included in the study. 42 on October 2, 2021 by guest. Protected copyright. 43 In cases where information from normal imaging preceding abnormal imaging results was 44 45 available we were certain of a window in time during which the haemorrhage did occur. In 46 47 most subjects we did not have the benefit of initial normal imaging. Here we used instead well 48 49 9 50 recognized imaging principles in judging the age of haemorrhage from its appearance on CT 51 52 or MR-imaging and estimated time of onset of bleed from this assessment. According to the 53 54 classification commonly used in the study of causes of cerebral palsy we classified the 55 56 haemorrhages as either intraventricular (IVH) and/or periventricular (PVH) or as 57 58 59 60 5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 10 3 parenchymal, in brain parenchyma not associated with the central ventricular system. The 4 5 neuroradiological information of likely time of onset was matched to all other clinical 6 7 information available. 8 9 10 In cases where the fetus or neonate died, autopsy reports were retrieved and studied to 11 12 evaluate whether ICH diagnosis was certain or unlikely. All reports from post-mortem 13 14 examinations were evaluated by an experienced perinatal pathologist (NP). Only cases 15 For peer review only 16 evaluated to be certain ICH were included. 17 18 19 20 21 Laboratory data 22 23 All laboratory data except maternal anti-HPA-1a antibody levels were collected from the 24 25 NOICH registry database (www.NOICH.org). 26 27 28 Maternal anti-HPA-1a antibody levels, except for the Finnish cases (four pregnancies), were 29 30 31 measured at the National reference laboratory of clinical platelet immunology in Tromsø,

32 11 33 Norway, by quantitative MAIPA. Reproducibility between Norwegian and Finnish

34 http://bmjopen.bmj.com/ 35 quantitation was secured by double analysis of some sera samples in both Norway and 36 37 Finland. In cases where several anti-HPA-1a antibody level measurements were available, the 38 39 highest maternal anti-HPA-1a antibody level measured during pregnancy or postpartum was 40 41 42 included in the study. on October 2, 2021 by guest. Protected copyright. 43 44 45 The lowest platelet count recorded in the fetus or new-born before any platelet transfusions 46 47 were given, was included. 48 49 50 51 52 Clinical data 53 54 Clinical data was mainly collected from the NOICH registry database (www.NOICH.org). 55 56 Additional clinical information was retrieved from the original medical records by each 57 58 country coordinator. 59 60 6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 The pregnancy where ICH was detected for the first time was referred to as the index case. A 4 5 subsequent ICH event was defined if ICH was recorded in any subsequent pregnancies after 6 7 the index ICH case. 8 9 10 11 12 Statistics 13 14 All data were analysed using SPSS software (Version 18.0 SPSS Inc., Chicago, IL). P<0.05 15 For peer review only 16 was considered significant. Means with 95% confidence intervals (CI) or median with range 17 18 were calculated for all continuous variables. Independent sample t-test and Kruskal-Wallis 19 20 21 test were used as appropriate. Oneway ANOVA test was used to test clinical outcome in 22 23 relation to time of bleeding onset. 24 25 26 27 Results 28 29 30 From the NOICH registry recording 592 FNAIT cases, 66 pregnancies from 57 women were 31 32 initially identified with suspected ICH. In 23 cases ICH or FNAIT diagnosis could not be 33

34 confirmed. In total, 43 confirmed cases of ICH due to FNAIT were recorded from 37 mothers. http://bmjopen.bmj.com/ 35 36 The 43 cases studied were included from The Netherlands, Finland, Sweden, Norway and 37 38 United Kingdom (Fig. 1). 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 Laboratory characteristics of ICH cases 44 45 HPA-1a alloimmunization was found to be the cause in 39 of 43 ICH cases (91%). Anti-HPA- 46 47 5b antibodies were found in two further cases. In one woman with records of two ICH cases, 48 49 50 incompatibility in the gpIa/IIa system was confirmed and in one of her ICH pregnancies anti- 51 52 HPA-5 antibodies were detected. 53 54 Data on maternal anti-HPA-1a antibody levels during pregnancy or after delivery were 55 56 available for 15 (35%) of the FNAIT pregnancies complicated by ICH. The median (range) 57 58 59 60 7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 highest anti-HPA-1a antibody level was 105 IU/mL (39 to 264 IU/mL). In two pregnancies 4 5 we have serial anti-HPA-1a antibody level measurements starting in 1st trimester, and in both 6 7 these pregnancies the anti-HPA-1a antibody levels fell from the 1st to 3rd trimester. The ICH 8 9 10 occurred between 28 and 34 weeks in these pregnancies. The median anti-HPA-1a antibody 11 st nd rd 12 levels measured during 1 trimester, 2 trimester, 3 trimester and post partum did not differ 13 14 significantly from each other or from the overall median highest antibody level (data not 15 For peer review only 16 shown). 17 18 The median (range) lowest fetal/ neonatal platelet count for ICH cases was 8x109/L (1 to 27). 19 20 21 In comparison, median (range) neonatal platelet count for previous or subsequent pregnancies 22 9 23 where no ICH was detected, was 17x10 /L (1 to 199) and significantly higher when compared 24 25 with ICH cases (Kruskal-Wallis test, P=0.004). 26 27 28 29 30 Clinical characteristics of ICH pregnancies 31 32 Maternal characteristics for all ICH cases are shown in Table 1. In the group of the 37 index 33

34 cases, 26 (70%) cases were first born children. However, most mothers had been pregnant http://bmjopen.bmj.com/ 35 36 before they had their first child: Eight women had one or more 1st trimester miscarriages and 37 38 six women had one or more 2nd trimester losses. For two women, we lack data on gravida 39 40 41 status. The mothers were primigravidae in only 10/37 (27%) index cases. In total, ten mothers 42 on October 2, 2021 by guest. Protected copyright. 43 (23%) experienced one or more 2nd trimester miscarriages (altogether 20) before or after the 44 45 ICH case. 46 47 Antenatal treatment was given in 4/43 (9%) of the pregnancies complicated by fetal/ neonatal 48 49 50 ICH. In 3 cases IVIG was given, two of these fetuses received additional intrauterine platelet 51 52 transfusions. One mother received corticosteroids as single treatment from 6 weeks gestation 53 54 onwards, due to two previous 2nd trimester miscarriages (not identified as FNAIT related). For 55 56 57 58 59 60 8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 the 37 index cases, only one mother received antenatal treatment. In this case, IVIG treatment 4 5 was started from week 7 due to autoimmune thrombocytopenia in the mother. 6 7 Twenty out of 43 mothers with pregnancies complicated by fetal/ neonatal ICH were 8 9 10 delivered by caesarean section (CS); CS was performed in nine cases with abnormal fetal 11 12 cerebral ultrasound as main indication. 13 14 A high proportion of the pregnancies complicated by ICH were ended preterm: 29 out of 43 15 For peer review only 16 children (67%) were born before 37 weeks and 4 children were born before 28 weeks. 17 18 Abnormal US scan with detection of ICH was the main reason for premature delivery (Table 19 20 21 2). 22 23 Five (12%) of the infants died in utero and one (2%) died during labour. Neonatal 24 25 characteristics of the live-born children are shown in Table 3. Nine children died within the 26 27 first four days after delivery. Most of the survivors developed neurological sequelae: Eight 28 29 30 infants were diagnosed with cerebral palsy, ten were reported to be moderate/ severely 31 32 mentally retarded or severely disabled, seven neonates were reported to have epilepsy and 33

34 four were blind or with severely reduced vision. In addition, one case of autism and one case http://bmjopen.bmj.com/ 35 36 of impaired hearing were reported. More than one neurological complication was reported in 37 38 several cases. Five (14%) of surviving neonates with ICH were reported to be alive and well 39 40 41 at time of discharge after delivery, but data on long-term clinical outcome is missing (Fig. 1). 42 on October 2, 2021 by guest. Protected copyright. 43 The fetuses/ neonates were male in the majority (65%) of ICH cases. There was no significant 44 45 difference in maternal anti-HPA-1a antibody levels, birth weight, APGAR scores or platelet 46 47 counts when comparing boys and girls (Independent sample T-test, P>0.05). 48 49 th 50 Ten (23%) ICH neonates were below the 10 percentile for birth weight and defined as small 51 52 for gestational age (SGA) according to standard growth curves. All SGA cases except one 53 54 were boys. 55 56 57 58 59 60 9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Bleeding characteristics 4 5 Neuroradiological studies were used to review and confirm 19 ICH cases. The remaining 24 6 7 ICH cases were confirmed using descriptions of radiographic material by local radiologists. 8 9 10 The estimated time of bleeding onset for each ICH case is shown in Figure 2. The figure 11 12 demonstrates that many bleedings started early, and that in most cases several weeks passed 13 14 between bleeding onset and delivery. More than half (23/43, 54%) of the bleedings happened 15 For peer review only 16 around gestational week 28 or earlier. In 16 (70%) of the 23 cases with bleeding onset before 17 18 the 3rd trimester we do not know the exact time of bleeding onset, only that the bleeding 19 20 21 started no later than 28 weeks (Figure 2). Twenty-nine out of 43 bleedings (67%) started 22 23 before 34 gestational weeks. No cases of intrapartum ICH bleedings were confirmed. The 24 25 time of bleeding onset in the 10 primigravid cases were not different from the multigravid 26 27 ICH cases. 28 29 30 There was no difference in clinical outcome in relation to time of bleeding onset (data not 31 32 shown). However, it is noteworthy that perinatal death occurred in two girls (14%) where the 33

34 bleedings were found to have happened before 30 weeks. In boys, perinatal death occurred in http://bmjopen.bmj.com/ 35 36 13 (46%) cases with bleedings happening also at a later gestational age. 37 38 Intraparenchymal haemorrhages were found in 11 out of 13 cases where the bleeding occurred 39 40 41 during the third trimester or after delivery. Figure 3a illustrates intraparenchymal 42 on October 2, 2021 by guest. Protected copyright. 43 haemorrhage. 44 45 In five cases, multiple bleeding episodes were found. All second bleedings occurred after 33 46 47 weeks (range 33 to 37 weeks). These five cases were all due to HPA-1a alloimmunization. 48 49 50 Fifteen ICH cases were classified as intraparenchymal and 13 cases as IVH/ PVH (Figure 3b). 51 52 Five cases were classified as miscellaneous (Figure 3c), whereas 10 cases could not be 53 54 classified. Among cases where ICH was found to occur before 28 weeks, all but one case was 55 56 found to be IVH/PVH. 57 58 59 60 10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 4 5 Treatment and clinical outcome in subsequent pregnancies 6 7 ICH was detected in six (23%) subsequent pregnancies. One woman had records of ICH in 8 9 10 two subsequent pregnancies after the index ICH case. In most subsequent pregnancies (20/26, 11 12 77%), no ICH was found. 13 14 In 19/ 26 (73%) of subsequent pregnancies, the mothers received antenatal treatment (IVIG). 15 For peer review only 16 The intravenous immunoglobulin schedules varied greatly, with a median starting time at 18 17 18 weeks (range 16-35 weeks). In the six cases where ICH was detected in subsequent 19 20 21 pregnancies, three received IVIG treatment. However, in one of these cases IVIG treatment 22 23 was started after ICH was detected. Therefore, IVIG treatment failed to prevent ICH in two 24 25 (11%) out of 19 cases. These two treatment failures were from the same woman. It should be 26 27 commented that the obstetrical history of this woman is particularly severe with regards to 28 29 30 FNAIT complications. The ICH recurrence rate was therefore 11% in IVIG treated 31 5 32 pregnancies. Compared with historical data reporting 79% risk of ICH recurrence in FNAIT, 33

34 our data indicate that IVIG was effective in preventing ICH. http://bmjopen.bmj.com/ 35 36 In seven cases, no IVIG treatment was given during a subsequent pregnancy after ICH was 37 38 detected in a previous pregnancy. Five of these untreated pregnancies come from Norway, 39 40 41 where IVIG is not routinely given as treatment during HPA alloimmunized pregnancies. ICH 42 on October 2, 2021 by guest. Protected copyright. 43 occurred in three of these seven cases. 44 45 46 47 Discussion 48 49 50 The main findings of this study are that the majority of ICH bleedings occurred by the end of 51 52 the second trimester and that clinical outcome was devastating for most cases. The high 53 54 frequency of bleedings occurring before 28 weeks indicates that the fetus may be severely 55 56 affected already in the second trimester. 57 58 59 60 11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 This is the largest study to date on fetal/ neonatal ICH caused by FNAIT. For the first time, 4 5 time of bleeding onset was assessed using clinical information together with radiographic 6 7 imaging and autopsy reports. The in-depth study of both laboratory and clinical information 8 9 10 was done in close collaboration between obstetricians, immunologists, perinatal pathologists 11 12 and specialists in neuroradiology. Limitations of this study include that it is a retrospective 13 14 cohort study, being subject to confounding and information bias. A bias toward inclusion of 15 For peer review only 16 the more severe ICH phenotype may be considered. The ICH cases in this study were 17 18 collected from five different countries and therefore reflect several institutions’ clinical 19 20 21 experiences. There is obvious heterogeneity in antenatal treatment for FNAIT between these 22 23 countries. However, since most of the patients affected by foetal ICH did not receive antenatal 24 25 treatment, we consider the study population uniform and representative of a larger population. 26 27 Some earlier studies have suggested the onset of bleedings to be in the third rather than the 28 29 3,12 30 second trimester, and are in variance with this study. The judgment of the onset of cerebral 31 32 bleeding in this study was cautious, and the onset was probably in many cases even earlier 33

34 than we report. However, previous studies included too few cases to address this question in http://bmjopen.bmj.com/ 35 36 an adequate manner. Also, and more importantly, other studies report the gestational age 37 38 when the ICH was diagnosed, and did not assess when the bleeding may have occurred. In a 39 40 13 41 recent study by Bussel et al, antenatal management to prevent recurrence of ICH caused by 42 on October 2, 2021 by guest. Protected copyright. 43 FNAIT was studied. Gestational age at the time of ICH is reported in this study, but without 44 45 any data with regards to how the timing of ICH was assessed. These data are therefore 46 47 difficult to assess, but in support of our data they report that as many as 8/37 (22%) of ICH 48 49 50 cases in their study population occurred before 28 gestational weeks. Most of the ICH cases 51 52 occurred in boys, and the bleedings were more often lethal when the fetus was male. The 53 54 finding that 80% of the SGA neonates in our population were boys supports the recently 55 56 published observation that birth weight in boys, but not in girls, was associated with maternal 57 58 59 60 12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 14 3 anti-HPA-1a antibodies. Data on maternal anti-HPA-1a antibody levels in FNAIT 4 5 pregnancies complicated by fetal/ neonatal ICH have not been published before. 6 7 Fetal ICH due to FNAIT often occurred in the first child. Most ICH cases will therefore not 8 9 10 be recognized in time for treatment or prophylactic measures if we do not identify 11 12 pregnancies at risk before the onset of the bleeding. The high number of first-born children in 13 14 the study population may not necessarily mean that the risk of ICH caused by FNAIT is 15 For peer review only 16 genuinely higher in the first-born child. The distribution could be skewed towards nulliparous 17 18 women since these women may choose not to have more children due to high recurrence risk. 19 20 21 Further, most of these women received antenatal treatment during the subsequent pregnancy 22 23 thereby reducing the incidence of ICH in the younger siblings. Nevertheless, this finding 24 25 challenges the current management strategy where antenatal treatment is given in subsequent 26 27 pregnancies after FNAIT has been diagnosed in the first child. A majority of these children 28 29 30 will suffer from bleeding already in the second trimester or in the early part of the third 31 th 32 trimester. Possible interventions to reduce risk of ICH need to be introduced before the 20 33

34 week of gestation. In most children the ICH was either fatal or induced severe disabilities. http://bmjopen.bmj.com/ 35 36 Our findings therefore support the idea of identifying HPA-1bb mothers with anti-HPA-1a 37 38 antibodies in all pregnancies3,15 and working towards a prophylactic approach to prevent the 39 40 16,17 41 immune response against HPA-1a. The rate of prematurity and CS was high in our study. 42 on October 2, 2021 by guest. Protected copyright. 43 Many of these patients required delivery before term due to problems related to the fetal ICH 44 45 such as fetal distress, lack of fetal movements, abnormal ultrasound scan etc. Thus, the 46 47 present study does not indicate that FNAIT per se is associated with an increased risk for 48 49 50 prematurity. 51 52 There were no confirmed cases of ICH occurring intrapartum in this study, and only two 53 54 bleedings occurred after delivery. This could suggest that mode of delivery may not be so 55 56 important in the prevention of ICH. However, the high CS rate in this study population may 57 58 59 60 13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 have contributed to the low number of intrapartum bleedings. Whether or not delivery by CS 4 5 prevents ICH needs to be further addressed.3,18 The present study suggests that IVIG 6 7 treatment during pregnancy is protective in regard to ICH in most cases, which is in 8 9 13 10 accordance with previous studies. However, it is an open question whether IVIG would have 11 12 protected the first born child from ICH, or whether there is a genuine increased risk of ICH in 13 14 the first born child. Further, it remains to be established if there is also a milder phenotype of 15 For peer review only 16 ICH with discrete symptoms and better outcome. This question can only be addressed in 17 18 prospective studies including general screening and repeated fetal US examinations. The 19 20 21 maternal anti-HPA-1a antibody levels were extremely high among ICH cases in this study. 22 23 Maternal anti-HPA-1a antibody levels may therefore be useful to identify pregnancies at risk 24 25 for fetal ICH, but these findings need to be evaluated in larger prospective studies. Finally, 26 27 why boys seem to be more susceptible to maternal anti-HPA-1a antibodies is currently not 28 29 30 known and needs further investigation. 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41

42 on October 2, 2021 by guest. Protected copyright. 43 Figure legends 44 45 Figure 1 46 47 Flow diagram of the study population 48 49 50 51 Figure 2 52 53 54 Estimated time period for onset of ICH is shown for 41 cases of ICH caused by FNAIT. An 55 56 arrowhead to the left indicates that the earliest time of onset cannot be estimated, only that the 57 58 bleeding occurred before the gestational age indicated on the x-axis. 59 60 14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 4 5 6 7 Figure 3 8 9 10 a) CT scan shows a very large intraparenchymal haemorrhage with mass effect and occupying 11 12 most of the right frontal lobe in a term baby. Note a second but smaller haemorrhage on the 13 14 left side adjacent to the ventricle but separate from this. The attenuation of both haemorrhages 15 For peer review only 16 indicates that the haemorrhage is several weeks old at the time of imaging. We estimated the 17 18 right haemorrhage to be 4 - 6 weeks old and the left still a couple of weeks older. The 19 20 21 pregnancy was considered normal, and no IVIG treatment was given. A live boy was born 22 9 23 vaginally at 40 gestational weeks with petecchiae. The platelet count was 6x10 /L. 24 25 26 27 b) This CT shows marked ventriculomegaly, partly due to hydrocephalus, partly due to loss of 28 29 30 brain tissue in the left hemisphere. Note the very large ventricle on the left side occupying 31 32 most of the left hemicranium. The shape of the left lateral ventricle is irregular and that of 33

34 residual from a previous intraventricular haemorrhage with a paraventricular atrophic defect http://bmjopen.bmj.com/ 35 36 caused by an old periventricular haemorrhagic infarction. This pattern is pathognomonic for 37 38 this condition even though no actual blood can be detected. The interpretation of this image is 39 40 41 that of an intraventricular haemorrhage associated with a periventricular hemorrhagic 42 on October 2, 2021 by guest. Protected copyright. 43 infarction timed at 28 – 30 gestational weeks or earlier. Hydrocephalus was diagnosed 44 45 intrapartum because of breech presentation. A live boy was born at term with a platelet count 46 47 of 4x109/L. The child has severe cerebral palsy and a hydrocephalus shunt. 48 49 50 51 52 c) The CT scan shows multiple focal intraparenchymal haemorrhages throughout both 53 54 cerebral hemispheres. Most haemorrhages are quite small. Note also the extensive extracranial 55 56 subgaleal haemorrhage overlying the right hemicranium. All bleedings are of same age and 57 58 59 60 15 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 maximum seven days old. It was noted that the fetus was small on US scan at 25 weeks, but 4 5 otherwise the pregnancy was considered normal. At 42 weeks a live boy was born vaginally, 6 7 with multiple petecchiaes and multiple retinal bleedings. The platelet count at delivery was 8 9 9 9 10 14x10 /L, with nadir value 8x10 /L. He received platelet transfusions. At time of discharge 11 12 the boy was described as alive and well, but we do not have data on long-term clinical 13 14 outcome. 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38

39 40 41 Table 1 Maternal characteristics of ICH-affected pregnancies 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 Maternal characteristic 46 47 Maternal age in years, mean (95% CI) 29.3 (27.6-31.0) 48 49 50 Obstetrical history 51 52 Primigravida, n (%) 10 (23) 53 54 First-born child, n (%) 27 (63) 55 56 2nd trimester miscarriage, n (%) 10 (23) 57 58 59 60 16 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Pre-eclampsia in this pregnancy, n (%) 3 (7) 4 5 Vaginal delivery of ICH neonate, n (%) 22* (51) 6 7 Caesarean section of ICH neonate, n (%) 20* (47) 8 9 10 Fetal/ maternal treatment 11 12 IVIG, n (%) 4 (9) 13 14 Steroids, n (%) 1 (2) 15 For peer review only 16 Intrauterine platelet transfusion, n (%) 3 (7) 17 18 19 No treatment 36 (84) 20 21 *Mode of delivery was not known for one ICH pregnancy 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 Table 2 Preterm birth 42 on October 2, 2021 by guest. Protected copyright. 43 Reason for preterm delivery Number of pregnancies (% of total) 44 45 Abnormal US scan (ICH detected) 15* (52) 46 47 48 Intrauterine fetal demise 5* (17) 49 50 Intrauterine growth restriction/ pre-eclampsia 3 (10) 51 52 Spontaneous preterm birth 4 (14) 53 54 55 Not known 2 (7) 56 57 Total 29** (100) 58 59 60 17 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 *One case was a complication during cordocentesis 4 **Including one twin pregnancy 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 Table 3 Clinical outcome characteristics of ICH cases 42 on October 2, 2021 by guest. Protected copyright. 43 Characteristic Result 44 45 All ICH cases 46 47 Gestational age at delivery, median (range) in weeks 35 (23-42) 48 49 50 Birth weight, mean (SD) in grams 2274 (832) 51 52 Sex, female/ male 14/28 53 54 55 56 57 58 59 60 18 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 4 Stillborn 6 (14) 5 6 7 Live-born children 37 (86) 8 9 10 APGAR score <7 after 5 minutes, n (%) 8 (23) 11 12 9 13 Lowest platelet count (range) x10 /L* 8.9 (1-27) 14 15 For peer review only 16 Neonatal death, n (%) 9 (21) 17 18 19 Alive and well at discharge, n (%) 5 (12) 20 21 22 Alive with neurological sequelae, n (%) 23 (53) 23 24 *Data available for 32 pregnancies 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37

38 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 Acknowledgements 47 48 49 We thank Dr. Eline S. Van den Akker, MD, PhD, Amsterdam, the Netherlands for her initial 50 51 work setting up the NOICH database and registry. No compensation was received for this 52 53 contribution. 54 55 56 57 58 59 60 19 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Contributors 4 5 6 MW, AH and HT designed the study. HT, AH and MW carried out the data analyses. HT 7 8 drafted the manuscript. All authors commented on the drafts of the manuscript and 9 10 11 contributed to the final version. AH and MW are guarantors of the study. All authors, external 12 13 and internal, had full access to all of the data (including statistical reports and tables) in the 14 15 study and canFor take responsibility peer for the review integrity of the data only and the accuracy of the data 16 17 analysis. 18 19 20 21 22 23 24 Funding 25 26 ALD have support from UCLH/UCL by funding from the Department of Health’s NIHR 27 28 Biomedical Research Centres funding scheme for the submitted work. The funder had no part 29 30 31 in the design or execution of the study or the analysis and interpretation of the results. 32 33

34 http://bmjopen.bmj.com/ 35 Conflicts of interest 36 37 All authors have completed the Unified Competing Interest form at 38 39 40 www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and 41 42 declare that (1) no support from any organisation for the submitted work; (2) AH has stocks on October 2, 2021 by guest. Protected copyright. 43 44 in Prophylix Pharma AS trying to develop a prophylaxis against HPA-1a immunization that 45 46 might have an interest in the submitted work in the previous 3 years; (3) their spouses, 47 48 49 partners, or children have no financial relationships that may be relevant to the submitted 50 51 work; and (4) none of the authors have non-financial interests that may be relevant to the 52 53 submitted work. 54 55 56 57 Details of ethics approval 58 59 60 20 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 All ethics approvals were obtained on a National level, based on local regulations: Sweden, 4 5 “Regionala etikprövningsnämnden”, approval number 2005/2:1. Norway, the Regional 6 7 Committee for Medical Research Ethics, North Norway, approval number 2006/1585. UK: 8 9 10 The study was done as part of institutional performance, and it received institutional review 11 12 board exemption. The Netherlands: Ethics approval was waived by the IRB of the LUMC. 13 14 Finland: The study was based on the NAIT registry kept by the Finnish Red Cross Blood 15 For peer review only 16 Service according to the Finnish law. 17 18 19 20 21 22 Copyright 23 24 25 The corresponding author has the right to grant on behalf of all authors and does grant on 26 27 behalf of all authors, an exclusive licence (or non exclusive for government employees) on a 28 29 30 worldwide basis to the BMJ Publishing Group Ltd to permit this article to be published in 31 32 BMJ editions and any other BMJPGL products and sublicences such use and exploit all 33

34 subsidiary rights, as set out in our licence. http://bmjopen.bmj.com/ 35 36 37 Data sharing 38 39 40 41 No additional data available 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 Reference List 48 49 50 (1) Mao C, Guo J, Chituwo BM. Intraventricular haemorrhage and its prognosis, prevention and treatment 51 in term infants. J Trop Pediatr 1999; 45(4):237-240. 52 53 (2) Williamson LM, Hackett G, Rennie J et al. The natural history of fetomaternal alloimmunization to the 54 platelet-specific antigen HPA-1a (PlA1, Zwa) as determined by antenatal screening. Blood 1998; 55 92:2280-2287. 56 57 (3) Kamphuis MM, Paridaans N, Porcelijn L et al. Screening in pregnancy for fetal or neonatal alloimmune 58 thrombocytopenia: systematic review. BJOG 2010; 117(11):1335-1343. 59 60 21 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 (4) Jocelyn LJ, Casiro OG. Neurodevelopmental outcome of term infants with intraventricular hemorrhage. 4 Am J Dis Child 1992; 146(2):194-197. 5 6 (5) Radder CM, Brand A, Kanhai HH. Will it ever be possible to balance the risk of intracranial haemorrhage in fetal or neonatal alloimmune thrombocytopenia against the risk of treatment strategies 7 to prevent it? Vox Sang 2003; 84(4):318-325. 8 9 (6) Kamphuis MM, Oepkes D. Fetal and neonatal alloimmune thrombocytopenia: prenatal interventions. 10 Prenat Diagn 2011; 31(7):712-719. 11 12 (7) Kanhai HH, Porcelijn L, Engelfriet CP et al. Management of alloimmune thrombocytopenia. Vox Sang 13 2007; 93(4):370-385. 14 15 (8) PachecoFor LD, Berkowitz peer RL, Moise KJ etreview al. Fetal and neonatal alloimmune only thrombocytopenia. A 16 management algorithm based on risk stratification. Obstet Gynecol 2012; 118(5):1157-1163. 17 18 (9) Osborn AG. Imaging of intracranial hemorrhage. Diagnostic neuroradiology. Mosby; 1994. 158-173. 19 20 (10) Bax M, tydeman C, Flodmark O. Clinical and MRI correlates of cerebral palsy. JAMA 2012; 21 296(13):1602-1608. 22 (11) Kiefel V, Santoso S, Weisheit M et al. Monoclonal antibody--specific immobilization of platelet 23 antigens (MAIPA): a new tool for the identification of platelet-reactive antibodies. Blood 1987; 24 70:1722-1726. 25 26 (12) Spencer JA, Burrows RF. Feto-maternal alloimmune thrombocytopenia: a literature review and 27 statistical analysis. Aust N Z J Obstet Gynaecol 2001; 41:45-55. 28 29 (13) Bussel JB, Berkowitz RL, Hung C, et al. Intracranial hemorrhage in alloimmune thrombocytopenia: 30 stratified management to prevent recurrence in the subsequent affected fetus. Am J Obstet Gynecol 31 2010; 203(2):135-14. 32 33 (14) Tiller H, Killie MK, Skogen B et al. Platelet antibodies and fetal growth: Maternal antibodies against fetal platelet antigen 1a are strongly associated with reduced birthweight in boys. Acta obstetricia et 34 http://bmjopen.bmj.com/ 35 gynecologica Scandinavica 2011; 91:79-86. 36 37 (15) Kjeldsen-Kragh J, Killie MK, Tomter G et al. A screening and intervention program aimed to reduce mortality and serious morbidity associated with severe neonatal alloimmune thrombocytopenia. Blood 38 2007; 110(3):833-839. 39 40 (16) Kjeldsen-Kragh J, Ni H, Skogen B. Towards a prophylactic treatment of HPA-related foetal and 41 neonatal alloimmune thrombocytopenia. Curr Opin Hematol 2012; 19(6):469-474. 42 on October 2, 2021 by guest. Protected copyright. 43 (17) Tiller H, Killie MK, Chen P et al. Toward a prophylaxis against fetal and neonatal alloimmune 44 thrombocytopenia: induction of antibody-mediated immune suppression and prevention of severe 45 clinical complications in a murine model. Transfusion 2012; 52(7):1446-1457. 46 47 (18) van den AE, Oepkes D, Brand A, et al. Vaginal delivery for fetuses at risk of alloimmune 48 thrombocytopenia? BJOG 2006; 113(7):781-783. 49 50 51

53 54 55 56 57 58 59 60 22 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Fetal intracranial haemorrhages caused by fetal and neonatal alloimmune 4 5 thrombocytopenia: An observational cohort study of 43 cases from an international 6 multicentre registry 7 1 2 3 8 Heidi Tiller HT, Marije M. Kamphuis MMK, Olof Flodmark OF, Nikos Papadogiannakis 9 4 5 6 6 6 10 NP, Anna L David ALD, Susanna Sainio SS, Sinikka Koskinen SK, Kaija Javela KJ, 11 Agneta Wikman AW,7 Riitta Kekomaki RK,6 Humphrey H.H. Kanhai HHHK,2 Dick Oepkes 12 2 10 11 13 DO, Anne Husebekk AH, * and Magnus Westgren MW * 14 *Corresponding authors 15 For peer review only 16 17 1. Department of Obstetrics and Gynaecology, University Hospital North Norway, 9038 18 Tromsø, Norway 19 20 2. Department of Obstetrics, Leiden University Medical Center, 2253LZ Leiden, The 21 Netherlands 22 3. Neuroradiologiska kliniken, Karolinska Universitetssjukhuset, 17176 Stockholm, 23 Sweden 24 4. Department of Laboratory Medicine, Division of Pathology, Section of Perinatal 25 Pathology, Karolinska Institutet, 14186 Stockholm, Sweden 26 5. Obstetrics and Maternal Foetal Medicine Institute for Women's Health, University 27 College, WC1E 6HX London, United Kingdom 28 6. Finnish Red Cross Blood Service, 00310 Helsinki, Finland 29 30 7. Clinical Immunology and Transfusion Medicine, Karolinska University Hospital, 31 14186 Stockholm, Sweden 32 8. Dept of Laboratory Medicine, Karolinska Institutet, 17176 Stockholm, Sweden 33 9. Department of Laboratory Medicine, University Hospital North Norway, 9038

44 45 46 47 48 49 50 51 52 Word count: 3609 53 54 55 56 57 58 59 60 1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Abstract 4 5 Objective To characterize pregnancies where the fetus or neonate was diagnosed with fetal 6 7 and neonatal alloimmune thrombocytopenia (FNAIT) and suffered from intracranial 8 9 10 haemorrhage (ICH), with special focus on time of bleeding onset. 11 12 Design Observational cohort study of all recorded cases of ICH caused by FNAIT from the 13 14 international No IntraCranial Haemorrhage (NOICH) registry during the period 2001-2010. 15 For peer review only 16 Setting Thirteen tertiary referral centres from nine countries across the world. 17 18 Participants 37 mothers and 43 children of FNAIT pregnancies complicated by fetal or 19 20 21 neonatal ICH identified from the NOICH registry was included if FNAIT diagnosis and ICH 22 23 was confirmed. 24 25 Primary and secondary outcome measures Gestational age at onset of ICH, type of ICH 26 27 and clinical outcome of ICH were the primary outcome measures. General maternal and 28 29 30 neonatal characteristics of pregnancies complicated by fetal/ neonatal ICH were secondary 31 32 outcome measures. 33

34 Results From a total of 592 FNAIT cases in the registry, 43 confirmed cases of ICH due to http://bmjopen.bmj.com/ 35 36 FNAIT were included in the study. The majority of bleedings (23/43, 54%) occurred before 37 38 28 gestational weeks and often affected the first born child (27/43, 63%). One third (35%) of 39 40 41 the children died within 4 days after delivery. 23 (53%) children survived with severe 42 on October 2, 2021 by guest. Protected copyright. 43 neurological disabilities and only five (12%) were alive and well at time of discharge. 44 45 Antenatal treatment was not given in most (91%) cases of fetal/ neonatal ICH. 46 47 Conclusions ICH caused by FNAIT often occurs during 2nd trimester and the clinical 48 49 50 outcome is poor. In order to prevent ICH caused by FNAIT, at risk pregnancies must be 51 nd 52 identified and prevention and/or interventions should start early in the 2 trimester. 53 54 Word count: 267 55 56 57 58 59 60 2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Article summary 4 5 Article focus 6 7 • Fetal and neonatal alloimmune thrombocytopenia (FNAIT) is the most common cause 8 9 10 of intracranial haemorrhage (ICH) in relation to thrombocytopenia in term born 11 12 infants. Gestational age at time of bleeding onset is important when antenatal 13 14 treatment options are discussed. No study has specifically addressed this before. 15 For peer review only 16 • From an international registry of almost 600 FNAIT cases, all recorded cases of ICH 17 18 19 were studied in detail. The aim of the study was to characterize pregnancies where the 20 21 fetus or neonate suffered from ICH caused by FNAIT, with special focus on 22 23 gestational age at time of bleeding onset. 24 25 Key messages 26 27 28 • The majority of ICH bleedings occurred by the end of the second trimester and clinical 29 30 outcome was devastating for most cases. Possible interventions to reduce risk of ICH 31 32 therefore need to be introduced before the 20th week of gestation. 33

34 • The majority ICH cases occurred in the first-born child. Most ICH cases will therefore http://bmjopen.bmj.com/ 35 36 37 not be recognized in time for treatment if we do not identify pregnancies at risk before 38 39 the first child is born. 40 41 • IVIG treatment during the subsequent pregnancy seemed to be protective with regard 42 on October 2, 2021 by guest. Protected copyright. 43 to ICH in most cases, reducing the ICH recurrence risk from 79% as previously 44 45 reported, to 11%. 46 47 48 Strengths and limitations 49 50 This is the largest study to date on ICH due to FNAIT. For the first time, gestational age at 51 52 bleeding onset was assessed in depth. Limitations of this study include that it is a 53 54 retrospective study, being subject to confounding and information bias. 55 56 57 58 59 60 3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Introduction 4 5 Fetal and neonatal alloimmune thrombocytopenia (FNAIT) is the most common cause of 6 1 7 intracranial haemorrhage (ICH) in relation to thrombocytopenia in term born infants. FNAIT 8 9 is caused by maternal alloantibodies directed against fetal platelets due to incompatibility in 10 11 human platelet antigens (HPAs). ICH due to FNAIT is reported to occur in 1: 12,500 – 25,000 12 13 births.2,3 The clinical outcome is often more severe than for neonatal ICH from other 14 15 1,4 For peer review only 5 16 causes. The ICH recurrence rate in subsequent pregnancies is reported to be 79%. 17 18 Therapies have been developed that can reduce the incidence of ICH, such as weekly high 19 20 dose intravenous immunoglobulin (IVIG).6,7,8 It is therefore considered important to identify 21 22 ICH caused by FNAIT in order to treat the mother in subsequent pregnancies. Gestational age 23 24 25 at time of bleeding onset is a key factor when antenatal treatment options are discussed since 26 27 treatment to prevent ICH needs to be commenced before ICH is likely to occur. No study has 28 29 specifically addressed this. Identifying common denominators of human HPA-alloimmunized 30 31 pregnancies complicated by ICH may serve as a future tool to help identify pregnancies at 32 33 high risk.

34 http://bmjopen.bmj.com/ 35 36 The No IntraCranial Haemorrhage (NOICH) registry is a multinational registry including 592 37 38 pregnancies complicated by FNAIT from 13 tertiary referral centres across the world. This 39 40 study presents the results of an in-depth evaluation of all recorded cases of ICH caused by 41 42 FNAIT from this registry. The aim of the study was to characterize pregnancies where the on October 2, 2021 by guest. Protected copyright. 43 44 45 fetus or neonate suffered from ICH with special focus on clinical and laboratory 46 47 characteristics and time of bleeding onset. 48 49 50 51 Materials and methods 52 53 Study design and inclusion criteria 54 55 56 All centres that had registered FNAIT cases in the NOICH registry from 2001-2010 were 57 58 invited to participate in this observational cohort study. Pregnancies recorded in the NOICH 59 60 4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 registry as complicated by fetal or neonatal ICH were identified, and included if both the 4 5 diagnosis of FNAIT and ICH were confirmed. STROBE guidelines were followed as 6 7 appropriate. 8 9 10 A case was defined as FNAIT and included if: 1) Incompatibility between maternal and 11 12 paternal/ fetal HPA type was confirmed and maternal anti-HPA antibodies were detected, 2) 13 14 HPA-incompatibility between the mother and father was confirmed and the fetus/neonate 15 For peer review only 16 suffered ICH and 3) anti-HPA antibodies were detected in the mother but data on fetal/ 17 18 paternal HPA genotype was missing. 19 20 21 Neuroradiological images were recovered and reviewed as electronic copies in a picture 22 23 archiving and communication system (PACS) using standard display programs. One 24 25 Norwegian subject and two of the Dutch subjects had MR-studies performed in utero. All 26 27 available neuroradiological images were re-evaluated for this study by an experienced 28 29 30 independent paediatric neuroradiologist (OF). The focus of the neuroradiological evaluation 31 32 was primarily to confirm the ICH diagnosis, secondly to use imaging to assist in assessing 33

34 time of bleeding onset and finally to classify the type of bleeding. When images were not http://bmjopen.bmj.com/ 35 36 available, written reports of the imaging evaluations for the patient files by others were used 37 38 to evaluate if the ICH diagnosis was correct. and theCases where an ICH could not be 39 40 41 confirmed subject could bewere not included in the study. 42 on October 2, 2021 by guest. Protected copyright. 43 In cases where information from normal imaging preceding abnormal imaging results was 44 45 available we were certain of a window in time during which the haemorrhage did occur. In 46 47 most subjects we did not have the benefit of initial normal imaging. Here we used instead well 48 49 9 50 recognized imaging principles in judging the age of haemorrhage from its appearance on CT 51 52 or MR-imaging and estimated time of onset of bleed from this assessment. According to the 53 54 classification commonly used in the study of causes of cerebral palsy we classified the 55 56 haemorrhages as either intraventricular (IVH) and/or periventricular (PVH) or as 57 58 59 60 5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 10 3 parenchymal, in brain parenchyma not associated with the central ventricular system. The 4 5 neuroradiological information of likely time of onset was matched to all other clinical 6 7 information available. 8 9 10 In cases where the fetus or neonate died, autopsy reports were retrieved and studied to 11 12 evaluate whether ICH diagnosis was certain or unlikely. All reports from post-mortem 13 14 examinations were evaluated by an experienced perinatal pathologist (NP). Only cases 15 For peer review only 16 evaluated to be certain ICH were included. 17 18 19 20 21 Laboratory data 22 23 All laboratory data except maternal anti-HPA-1a antibody levels were collected from the 24 25 NOICH registry database (www.NOICH.org). 26 27 28 Maternal anti-HPA-1a antibody levels, except for the Finnish cases (four pregnancies), were 29 30 31 measured at the National reference laboratory of clinical platelet immunology in Tromsø,

32 11 33 Norway, by quantitative MAIPA. Reproducibility between Norwegian and Finnish

34 http://bmjopen.bmj.com/ 35 quantitation was secured by double analysis of some sera samples in both Norway and 36 37 Finland. In cases where several anti-HPA-1a antibody level measurements were available, the 38 39 highest maternal anti-HPA-1a antibody level measured during pregnancy or postpartum was 40 41 42 included in the study. on October 2, 2021 by guest. Protected copyright. 43 44 45 The lowest platelet count recorded in the fetus or new-born before any platelet transfusions 46 47 were given, was included. 48 49 50 51 52 Clinical data 53 54 Clinical data was mainly collected from the NOICH registry database (www.NOICH.org). 55 56 Additional clinical information was retrieved from the original medical records by each 57 58 country coordinator. 59 60 6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 The pregnancy where ICH was detected for the first time was referred to as the index case. A 4 5 subsequent ICH event was defined if ICH was recorded in any subsequent pregnancies after 6 7 the index ICH case. 8 9 10 11 12 Statistics 13 14 All data were analysed using SPSS software (Version 18.0 SPSS Inc., Chicago, IL). P<0.05 15 For peer review only 16 was considered significant. Means with 95% confidence intervals (CI) or median with range 17 18 were calculated for all continuous variables. Independent sample t-test and Kruskal-Wallis 19 20 21 test were used as appropriate. Oneway ANOVA test was used to test clinical outcome in 22 23 relation to time of bleeding onset. 24 25 26 27 Results 28 29 30 From the NOICH registry recording 592 FNAIT cases, 66 pregnancies from 57 women were 31 32 initially identified with suspected ICH. In 23 cases ICH or FNAIT diagnosis could not be 33

34 confirmed. In total, 43 confirmed cases of ICH due to FNAIT were recorded from 37 mothers. http://bmjopen.bmj.com/ 35 36 The 43 cases studied were included from The Netherlands, Finland, Sweden, Norway and 37 38 United Kingdom (Fig. 1). 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 Laboratory characteristics of ICH cases 44 45 HPA-1a alloimmunization was found to be the cause in 39 of 43 ICH cases (91%). Anti-HPA- 46 47 5b antibodies were found in two further cases. In one woman with records of two ICH cases, 48 49 50 incompatibility in the gpIa/IIa system was confirmed and in one of her ICH pregnancies anti- 51 52 HPA-5a and-5b antibodies were detected. 53 54 Data on maternal anti-HPA-1a antibody levels during pregnancy or after delivery were 55 56 available for 15 (35%) of the FNAIT pregnancies complicated by ICH. The median (range) 57 58 59 60 7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 highest anti-HPA-1a antibody level was 105 IU/mL (39 to 264 IU/mL). In two pregnancies 4 5 we have serial anti-HPA-1a antibody level measurements starting in 1st trimester, and in both 6 7 these pregnancies the anti-HPA-1a antibody levels fell from the 1st to 3rd trimester. The ICH 8 9 10 occurred between 28 and 34 weeks in these pregnancies. The median anti-HPA-1a antibody 11 st nd rd 12 levels measured during 1 trimester, 2 trimester, 3 trimester and post partum did not differ 13 14 significantly from each other or from the overall median highest antibody level (data not 15 For peer review only 16 shown). 17 18 The median (range) lowest fetal/ neonatal platelet count for ICH cases was 8x109/L (1 to 27). 19 20 21 In comparison, median (range) neonatal platelet count for previous or subsequent pregnancies 22 9 23 where no ICH was detected, was 17x10 /L (1 to 199) and significantly higher when compared 24 25 with ICH cases (Kruskal-Wallis test, P=0.004). 26 27 28 29 30 Clinical characteristics of ICH pregnancies 31 32 Maternal characteristics for all ICH cases are shown in Table 1. In the group of the 37 index 33

34 cases, 26 (70%) cases were first born children. However, most mothers had been pregnant http://bmjopen.bmj.com/ 35 36 before they had their first child: the mothers were primigravidae in only 10/37 (27%) index 37 38 cases. Eight women had one or more 1st trimester miscarriages and six women had one or 39 40 nd 41 more 2 trimester losses. For two women, we lack data on gravida status. The mothers were 42 on October 2, 2021 by guest. Protected copyright. 43 primigravidae in only 10/37 (27%) index cases.Te In total, tn en mothers (23%) experienced 44 45 one or more 2nd trimester miscarriages (altogether 20) before or after the ICH case. 46 47 Antenatal treatment was given in 4/43 (9%) of the pregnancies complicated by fetal/ neonatal 48 49 50 ICH. In 3 cases IVIG was given, two of these fetuses received additional intrauterine platelet 51 52 transfusions. One mother received corticosteroids as single treatment from 6 weeks gestation 53 54 onwards, due to two previous 2nd trimester miscarriages (not identified as FNAIT related). For 55 56 57 58 59 60 8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 the 37 index cases, only one mother received antenatal treatment. In this case, IVIG treatment 4 5 was started from week 7 due to autoimmune thrombocytopenia in the mother. 6 7 Twenty out of 43 mothers with pregnancies complicated by fetal/ neonatal ICH were 8 9 10 delivered by caesarean section (CS); CS was performed in nine cases with abnormal fetal 11 12 cerebral ultrasound as main indication. 13 14 A high proportion of the pregnancies complicated by ICH were ended preterm: 29 out of 43 15 For peer review only 16 children (67%) were born before 37 weeks and 4 children were born before 28 weeks. 17 18 Abnormal US scan with detection of ICH was the main reason for premature delivery (Table 19 20 21 2). 22 23 Five (12%) of the infants died in utero and one (2%) died during labour. Neonatal 24 25 characteristics of the live-born children are shown in Table 3. Nine children died within the 26 27 first four days after delivery. Most of the survivors developed neurological sequelae: Eight 28 29 30 infants were diagnosed with cerebral palsy, ten were reported to be moderate/ severely 31 32 mentally retarded or severely disabled, seven neonates were reported to have epilepsy and 33

34 four were blind or with severely reduced vision. In addition, one case of autism and one case http://bmjopen.bmj.com/ 35 36 of impaired hearing were reported. More than one neurological complication was reported in 37 38 several cases. Five (14%) of surviving neonates with ICH were reported to be alive and well 39 40 41 at time of discharge after delivery, but data on long-term clinical outcome is missing (.Fig. 1). 42 on October 2, 2021 by guest. Protected copyright. 43 The fetuses/ neonates were male in the majority (65%) of ICH cases. There was no significant 44 45 difference in maternal anti-HPA-1a antibody levels, birth weight, APGAR scores or platelet 46 47 counts when comparing boys and girls (Independent sample T-test, P>0.05). 48 49 th 50 Ten (23%) ICH neonates were below the 10 percentile for birth weight and defined as small 51 52 for gestational age (SGA) according to standard growth curves. All SGA cases except one 53 54 were boys. 55 56 57 58 59 60 9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Bleeding characteristics 4 5 Neuroradiological studies were used to review and confirm 19 ICH cases. The remaining 24 6 7 ICH cases were confirmed using descriptions of radiographic material by local radiologists. 8 9 10 The estimated time of bleeding onset for each ICH case is shown in Figure 2. The figure 11 12 demonstrates that many bleedings started early, and that in most cases several weeks passed 13 14 between bleeding onset and delivery. More than half (23/43, 54%) of the bleedings happened 15 For peer review only 16 around gestational week 28 or earlier. In 16 (70%) of the 23 cases with bleeding onset before 17 18 the 3rd trimester we do not know the exact time of bleeding onset, only that the bleeding 19 20 21 started no later than 28 weeks (Figure 2). Twenty-nine out of 43 bleedings (67%) started 22 23 before 34 gestational weeks. No cases of intrapartum ICH bleedings were confirmed. The 24 25 time of bleeding onset in the 10 primigravid cases were not different from the multigravid 26 27 ICH cases. 28 29 30 There was no difference in clinical outcome in relation to time of bleeding onset (Figure 2, 31 32 Oneway ANOVA, P=0.75data not shown). However, it is noteworthy that perinatal death 33

34 occurred in two girls (14%) where the bleedings were found to have happened before 30 http://bmjopen.bmj.com/ 35 36 weeks. In boys, perinatal death occurred in 13 (46%) cases with bleedings happening also at a 37 38 later gestational age (data not shown). 39 40 41 Intraparenchymal haemorrhages were found in 11 out of 13 cases where the bleeding occurred 42 on October 2, 2021 by guest. Protected copyright. 43 during the third trimester or after delivery. See Figure 3a for illustration ofillustrates 44 45 intraparenchymal haemorrhage. 46 47 In five cases, multiple bleeding episodes were found. All second bleedings occurred after 33 48 49 50 weeks (range 33 to 37 weeks). These five cases were all due to HPA-1a alloimmunization. 51 52 Fifteen ICH cases were classified as intraparenchymal and 13 cases as IVH/ PVH (Figure 3b). 53 54 Five cases were classified as miscellaneous (Figure 3c), whereas 10 cases could not be 55 56 57 58 59 60 10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 classified. Among cases where ICH was found to occur before 28 weeks, all but one case was 4 5 found to be IVH/PVH. 6 7 8 9 10 Treatment and clinical outcome in subsequent pregnancies 11 12 ICH was detected in six (23%) subsequent pregnancies. One woman had records of ICH in 13 14 two subsequent pregnancies after the index ICH case. In most subsequent pregnancies (20/26, 15 For peer review only 16 77%), no ICH was found. 17 18 In 19/ 26 (73%) of subsequent pregnancies, the mothers received antenatal treatment (IVIG). 19 20 21 The intravenous immunoglobulin schedules varied greatly, with a median starting time at 18 22 23 weeks (range 16-35 weeks). In the six cases where ICH was detected in subsequent 24 25 pregnancies, three received IVIG treatment. However, in one of these cases IVIG treatment 26 27 was started after ICH was detected. Therefore, IVIG treatment failed to prevent ICH in two 28 29 30 (11%) out of 19 cases. These two treatment failures were from the same woman. It should be 31 32 commented that the obstetrical history of this woman is particularly severe with regards to 33

34 FNAIT complications. The ICH recurrence rate was therefore 11% in IVIG treated http://bmjopen.bmj.com/ 35 36 pregnancies. Compared with historical data reporting 79% risk of ICH recurrence in FNAIT,5 37 38 our data indicate that IVIG was effective in preventing ICH. 39 40 41 In seven cases, no IVIG treatment was given during a subsequent pregnancy after ICH was 42 on October 2, 2021 by guest. Protected copyright. 43 detected in a previous pregnancy. Five of these untreated pregnancies come from Norway, 44 45 where IVIG is not routinely given as treatment during HPA alloimmunized pregnancies. ICH 46 47 occurred in three of these seven cases. 48 49 50 51 52 Discussion 53 54 The main findings of this study are that the majority of ICH bleedings occurred by the end of 55 56 the second trimester and that clinical outcome was devastating for most cases. The high 57 58 59 60 11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 frequency of bleedings occurring before 28 weeks indicates that the fetus may be severely 4 5 affected already in the second trimester. 6 7 This is the largest study to date on fetal/ neonatal ICH caused by FNAIT. For the first time, 8 9 10 time of bleeding onset was assessed using clinical information together with radiographic 11 12 imaging and autopsy reports. The in-depth study of both laboratory and clinical information 13 14 was done in close collaboration between obstetricians, immunologists, perinatal pathologists 15 For peer review only 16 and specialists in neuroradiology. Limitations of this study include that it is a retrospective 17 18 cohort study, being subject to confounding and information bias. A bias toward inclusion of 19 20 21 the more severe ICH phenotype may be considered. The ICH cases in this study were 22 23 collected from five different countries and therefore reflect several institutions’ clinical 24 25 experiences. There is obvious heterogeneity in antenatal treatment for FNAIT between these 26 27 countries. However, since most of the patients affected by foetal ICH did not receive antenatal 28 29 30 treatment, we consider the study population uniform and representative of a larger population. 31 32 Some earlier studies have suggested the onset of bleedings to be in the third rather than the 33 3,12

34 second trimester, and are in variance with this study. The judgment of the onset of cerebral http://bmjopen.bmj.com/ 35 36 bleeding in this study was cautious, and the onset was probably in many cases even earlier 37 38 than we report. However, previous studies included too few cases to address this question in 39 40 41 an adequate manner. Also, and more importantly, other studies report the gestational age 42 on October 2, 2021 by guest. Protected copyright. 43 when the ICH was diagnosed, and did not assess when the bleeding may have occurred. In a 44 45 recent study by Bussel et al,13 antenatal management to prevent recurrence of ICH caused by 46 47 FNAIT was studied. Gestational age at the time of ICH is reported in this study, but without 48 49 50 any data with regards to how the timing of ICH was assessed. These data are therefore 51 52 difficult to assess, but in support of our data they report that as many as 8/37 (22%) of ICH 53 54 cases in their study population occurred before 28 gestational weeks. Most of the ICH cases 55 56 occurred in boys, and the bleedings were more often lethal when the fetus was male. The 57 58 59 60 12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 35 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 finding that 80% of the SGA neonates in our population were boys supports the recently 4 5 published observation that birth weight in boys, but not in girls, was associated with maternal 6 7 anti-HPA-1a antibodies.14 Data on maternal anti-HPA-1a antibody levels in FNAIT 8 9 10 pregnancies complicated by fetal/ neonatal ICH have not been published before. 11 12 Fetal ICH due to FNAIT often occurred in the first child. Most ICH cases will therefore not 13 14 be recognized in time for treatment or prophylactic measures if we do not identify 15 For peer review only 16 pregnancies at risk before the onset of the bleeding. The high number of first-born children in 17 18 the study population may not necessarily mean that the risk of ICH caused by FNAIT is 19 20 21 genuinely higher in the first-born child. The distribution could be skewed towards nulliparous 22 23 women since these women may choose not to have more children due to high recurrence risk. 24 25 Further, most of these women received antenatal treatment during the subsequent pregnancy 26 27 thereby reducing the incidence of ICH in the younger siblings. TNevertheless, this finding 28 29 30 therefore challenges the current management strategy where antenatal treatment is given in 31 32 subsequent pregnancies after FNAIT has been diagnosed in the first child. A majority of these 33

34 children will suffer from bleeding already in the second trimester or in the early part of the http://bmjopen.bmj.com/ 35 36 third trimester. Possible interventions to reduce risk of ICH need to be introduced before the 37 38 20th week of gestation. In most children the ICH was either fatal or induced severe 39 40 41 disabilities. Our findings therefore support the idea of identifying HPA-1bb mothers with anti- 42 on October 2, 2021 by guest. Protected copyright. 43 HPA-1a antibodies in all pregnancies3,15 and working towards a prophylactic approach to 44 45 prevent the immune response against HPA-1a.16,17 The rate of prematurity and CS was high in 46 47 our study. Many of these patients required delivery before term due to problems related to the 48 49 50 fetal ICH such as fetal distress, lack of fetal movements, abnormal ultrasound scan etc. Thus, 51 52 the present study does not indicate that FNAIT per se is associated with an increased risk for 53 54 prematurity. 55 56 57 58 59 60 13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 36 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 There were no confirmed cases of ICH occurring intrapartum in this study, and only two 4 5 bleedings occurred after delivery. This could suggest that mode of delivery may not be so 6 7 important in the prevention of ICH. However, the high CS rate in this study population may 8 9 10 have contributed to the low number of intrapartum bleedings. Whether or not delivery by CS 11 3,18 12 prevents ICH needs to be further addressed. The present study suggests that IVIG 13 14 treatment during pregnancy is protective in regard to ICH in most cases, which is in 15 For peer review only 16 accordance with previous studies.13 However, it is an open question whether IVIG would have 17 18 protected the first born child from ICH, or whether there is a genuine increased risk of ICH in 19 20 21 the first born child. Further, it remains to be established if there is also a milder phenotype of 22 23 ICH with discrete symptoms and better outcome. This question can only be addressed in 24 25 prospective studies including general screening and repeated fetal US examinations. The 26 27 maternal anti-HPA-1a antibody levels were extremely high among ICH cases in this study. 28 29 30 Maternal anti-HPA-1a antibody levels may therefore be useful to identify pregnancies at risk 31 32 for fetal ICH, but these findings need to be evaluated in larger prospective studies. Finally, 33

34 why boys seem to be more susceptible to maternal anti-HPA-1a antibodies is currently not http://bmjopen.bmj.com/ 35 36 known and needs further investigation. 37 38 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 Figure legends 50 51 Figure 1 52 53 54 Flow diagram of the study population 55 56 57 58 Figure 2 59 60 14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Estimated time period for onset of ICH is shown for 41 cases of ICH caused by FNAIT. An 4 5 arrowhead to the left indicates that the earliest time of onset cannot be estimated, only that the 6 7 bleeding occurred before the gestational age indicated on the x-axis. 8 9 10 11 12 13 14 Figure 3 15 For peer review only 16 a) CT scan shows a very large intraparenchymal haemorrhage with mass effect and occupying 17 18 most of the right frontal lobe in a term baby. Note a second but smaller haemorrhage on the 19 20 21 left side adjacent to the ventricle but separate from this. The attenuation of both haemorrhages 22 23 indicates that the haemorrhage is several weeks old at the time of imaging. We estimated the 24 25 right haemorrhage to be 4 - 6 weeks old and the left still a couple of weeks older. The 26 27 pregnancy was considered normal, and no IVIG treatment was given. A live boy was born 28 29 9 30 vaginally at 40 gestational weeks with petecchiae. The platelet count was 6x10 /L. 31 32 33

34 b) This CT shows marked ventriculomegaly, partly due to hydrocephalus, partly due to loss of http://bmjopen.bmj.com/ 35 36 brain tissue in the left hemisphere. Note the very large ventricle on the left side occupying 37 38 most of the left hemicranium. The shape of the left lateral ventricle is irregular and that of 39 40 41 residual from a previous intraventricular haemorrhage with a paraventricular atrophic defect 42 on October 2, 2021 by guest. Protected copyright. 43 caused by an old periventricular haemorrhagic infarction. This pattern is pathognomonic for 44 45 this condition even though no actual blood can be detected. The interpretation of this image is 46 47 that of an intraventricular haemorrhage associated with a periventricular hemorrhagic 48 49 50 infarction timed at 28 – 30 gestational weeks or earlier. Hydrocephalus was diagnosed 51 52 intrapartum because of breech presentation. A live boy was born at term with a platelet count 53 54 of 4x109/L. The child has severe cerebral palsy and a hydrocephalus shunt. 55 56 57 58 59 60 15 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 38 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 c) The CT scan shows multiple focal intraparenchymal haemorrhages throughout both 4 5 cerebral hemispheres. Most haemorrhages are quite small. Note also the extensive extracranial 6 7 subgaleal haemorrhage overlying the right hemicranium. All bleedings are of same age and 8 9 10 maximum seven days old. It was noted that the fetus was small on US scan at 25 weeks, but 11 12 otherwise the pregnancy was considered normal. At 42 weeks a live boy was born vaginally, 13 14 with multiple petecchiaes and multiple retinal bleedings. The platelet count at delivery was 15 For peer review only 16 14x109/L, with nadir value 8x109/L. He received platelet transfusions. At time of discharge 17 18 the boy was described as alive and well, but we do not have data on long-term clinical 19 20 21 outcome. 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38

39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 Table 1 Maternal characteristics of ICH-affected pregnancies 48 49 50 51 52 Maternal characteristic 53 54 Maternal age in years, mean (95% CI) 29.3 (27.6-31.0) 55 56 Obstetrical history 57 58 59 60 16 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Primigravida, n (%) 10 (23) 4 5 First-born child, n (%) 27 (63) 6 7 2nd trimester miscarriage, n (%) 10 (23) 8 9 10 Pre-eclampsia in this pregnancy, n (%) 3 (7) 11 12 Vaginal delivery of ICH neonate, n (%) 22* (51) 13 14 Caesarean section of ICH neonate, n (%) 20* (47) 15 For peer review only 16 Fetal/ maternal treatment 17 18 19 IVIG, n (%) 4 (9) 20 21 Steroids, n (%) 1 (2) 22 23 Intrauterine platelet transfusion, n (%) 3 (7) 24 25 26 No treatment 36 (84) 27 28 *Mode of delivery was not known for one ICH pregnancy 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 48 Table 2 Preterm birth 49 50 Reason for preterm delivery Number of pregnancies (% of total) 51 52 Abnormal US scan (ICH detected) 15* (52) 53 54 55 Intrauterine fetal demise 5* (17) 56 57 Intrauterine growth restriction/ pre-eclampsia 3 (10) 58 59 60 17 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 40 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Spontaneous preterm birth 4 (14) 4 5 Not known 2 (7) 6 7 Total 29** (100) 8 9 10 *One case was a complication during cordocentesis 11 **Including one twin pregnancy 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 Table 3 Clinical outcome characteristics of ICH cases 48 49 50 Characteristic Result 51 52 All ICH cases 53 54 Gestational age at delivery, median (range) in weeks 35 (23-42) 55 56 Birth weight, mean (SD) in grams 2274 (832) 57 58 59 60 18 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 41 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Sex, female/ male 14/28 4 5 6 Stillborn 6 (14) 7 8 9 Live-born children 37 (86) 10 11 12 APGAR score <7 after 5 minutes, n (%) 8 (23) 13 14 15 Lowest plateletFor count peer(range) x109/L* review 8.9 (1-27)only 16 17 18 Neonatal death, n (%) 9 (21) 19 20 21 Alive and well at discharge, n (%) 5 (12) 22 23 24 Alive with neurological sequelae, n (%) 23 (53) 25 26 27 *Data available for 32 pregnancies 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 Acknowledgements 50 51 We thank Dr. Eline S. Van den Akker, MD, PhD, Amsterdam, the Netherlands for her initial 52 53 work setting up the NOICH database and registry. No compensation was received for this 54 55 contribution. 56 57 58 59 60 19 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 42 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 Contributors 4 5 6 MW, AH and HT designed the study. HT, AH and MW carried out the data analyses. HT 7 8 drafted the manuscript. All authors commented on the drafts of the manuscript and 9 10 11 contributed to the final version. AH and MW are guarantors of the study. All authors, external 12 13 and internal, had full access to all of the data (including statistical reports and tables) in the 14 15 study and canFor take responsibility peer for the review integrity of the data only and the accuracy of the data 16 17 analysis. 18 19 20 21 22 23 24 Funding 25 26 ALD have support from UCLH/UCL by funding from the Department of Health’s NIHR 27 28 Biomedical Research Centres funding scheme for the submitted work. The funder had no part 29 30 31 in the design or execution of the study or the analysis and interpretation of the results. 32 33

34 http://bmjopen.bmj.com/ 35 Conflicts of interest 36 37 All authors have completed the Unified Competing Interest form at 38 39 40 www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and 41 42 declare that (1) no support from any organisation for the submitted work; (2) AH has stocks on October 2, 2021 by guest. Protected copyright. 43 44 in Prophylix Pharma AS trying to develop a prophylaxis against HPA-1a immunization that 45 46 might have an interest in the submitted work in the previous 3 years; (3) their spouses, 47 48 49 partners, or children have no financial relationships that may be relevant to the submitted 50 51 work; and (4) none of the authors have non-financial interests that may be relevant to the 52 53 submitted work. 54 55 56 57 Details of ethics approval 58 59 60 20 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 43 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 All ethics approvals were obtained on a National level, based on local regulations: Sweden, 4 5 “Regionala etikprövningsnämnden”, approval number 2005/2:1. Norway, the Regional 6 7 Committee for Medical Research Ethics, North Norway, approval number 2006/1585. UK: 8 9 10 The study was done as part of institutional performance, and it received institutional review 11 12 board exemption. The Netherlands: Ethics approval was waived by the IRB of the LUMC. 13 14 Finland: The study was based on the NAIT registry kept by the Finnish Red Cross Blood 15 For peer review only 16 Service according to the Finnish law. 17 18 19 20 21 22 Copyright 23 24 25 The corresponding author has the right to grant on behalf of all authors and does grant on 26 27 behalf of all authors, an exclusive licence (or non exclusive for government employees) on a 28 29 30 worldwide basis to the BMJ Publishing Group Ltd to permit this article to be published in 31 32 BMJ editions and any other BMJPGL products and sublicences such use and exploit all 33

34 subsidiary rights, as set out in our licence. http://bmjopen.bmj.com/ 35 36 37 Data sharing 38 39 40 41 No additional data available 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 Reference List 48 49 50 (1) Mao C, Guo J, Chituwo BM. Intraventricular haemorrhage and its prognosis, prevention and treatment 51 in term infants. J Trop Pediatr 1999; 45(4):237-240. 52 53 (2) Williamson LM, Hackett G, Rennie J et al. The natural history of fetomaternal alloimmunization to the 54 platelet-specific antigen HPA-1a (PlA1, Zwa) as determined by antenatal screening. Blood 1998; 55 92:2280-2287. 56 57 (3) Kamphuis MM, Paridaans N, Porcelijn L et al. Screening in pregnancy for fetal or neonatal alloimmune 58 thrombocytopenia: systematic review. BJOG 2010; 117(11):1335-1343. 59 60 21 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 44 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 (4) Jocelyn LJ, Casiro OG. Neurodevelopmental outcome of term infants with intraventricular hemorrhage. 4 Am J Dis Child 1992; 146(2):194-197. 5 6 (5) Radder CM, Brand A, Kanhai HH. Will it ever be possible to balance the risk of intracranial haemorrhage in fetal or neonatal alloimmune thrombocytopenia against the risk of treatment strategies 7 to prevent it? Vox Sang 2003; 84(4):318-325. 8 9 (6) Kamphuis MM, Oepkes D. Fetal and neonatal alloimmune thrombocytopenia: prenatal interventions. 10 Prenat Diagn 2011; 31(7):712-719. 11 12 (7) Kanhai HH, Porcelijn L, Engelfriet CP et al. Management of alloimmune thrombocytopenia. Vox Sang 13 2007; 93(4):370-385. 14 15 (8) PachecoFor LD, Berkowitz peer RL, Moise KJ etreview al. Fetal and neonatal alloimmune only thrombocytopenia. A 16 management algorithm based on risk stratification. Obstet Gynecol 2012; 118(5):1157-1163. 17 18 (9) Osborn AG. Imaging of intracranial hemorrhage. Diagnostic neuroradiology. Mosby; 1994. 158-173. 19 20 (10) Bax M, tydeman C, Flodmark O. Clinical and MRI correlates of cerebral palsy. JAMA 2012; 21 296(13):1602-1608. 22 (11) Kiefel V, Santoso S, Weisheit M et al. Monoclonal antibody--specific immobilization of platelet 23 antigens (MAIPA): a new tool for the identification of platelet-reactive antibodies. Blood 1987; 24 70:1722-1726. 25 26 (12) Spencer JA, Burrows RF. Feto-maternal alloimmune thrombocytopenia: a literature review and 27 statistical analysis. Aust N Z J Obstet Gynaecol 2001; 41:45-55. 28 29 (13) Bussel JB, Berkowitz RL, Hung C, Kolb EA et al. Intracranial hemorrhage in alloimmune 30 thrombocytopenia: stratified management to prevent recurrence in the subsequent affected fetus. Am J 31 Obstet Gynecol 2010; 203(2):135-14. 32 33 (14) Tiller H, Killie MK, Skogen B et al. Platelet antibodies and fetal growth: Maternal antibodies against fetal platelet antigen 1a are strongly associated with reduced birthweight in boys. Acta obstetricia et 34 http://bmjopen.bmj.com/ 35 gynecologica Scandinavica 2011; 91:79-86. 36 37 (15) Kjeldsen-Kragh J, Killie MK, Tomter G et al. A screening and intervention program aimed to reduce mortality and serious morbidity associated with severe neonatal alloimmune thrombocytopenia. Blood 38 2007; 110(3):833-839. 39 40 (16) Kjeldsen-Kragh J, Ni H, Skogen B. Towards a prophylactic treatment of HPA-related foetal and 41 neonatal alloimmune thrombocytopenia. Curr Opin Hematol 2012; 19(6):469-474. 42 on October 2, 2021 by guest. Protected copyright. 43 (17) Tiller H, Killie MK, Chen P et al. Toward a prophylaxis against fetal and neonatal alloimmune 44 thrombocytopenia: induction of antibody-mediated immune suppression and prevention of severe 45 clinical complications in a murine model. Transfusion 2012; 52(7):1446-1457. 46 47 (18) van den AE, Oepkes D, Brand A, Kanhai HH. Vaginal delivery for fetuses at risk of alloimmune 48 thrombocytopenia? BJOG 2006; 113(7):781-783. 49 50 51

53 54 55 56 57 58 59 60 22 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 45 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 STROBE Statement—Checklist of items that should be included in reports of cohort studies 2 Item 3 No Recommendation 4 Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the abstract 5 6 Ok 7 (b) Provide in the abstract an informative and balanced summary of what was done 8 and what was found 9 Ok 10 11 Introduction 12 Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 13 Ok 14 15 Objectives For3 Statepeer specific objectives, review including any prespecified only hypotheses 16 Ok 17 Methods 18 Study design 4 Present key elements of study design early in the paper 19 20 Ok 21 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, 22 exposure, follow-up, and data collection 23 Ok 24 a 25 Participants 6 ( ) Give the eligibility criteria, and the sources and methods of selection of 26 participants. Describe methods of follow-up Ok 27 28 (b) For matched studies, give matching criteria and number of exposed and unexposed 29 NA 30 31 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and effect 32 modifiers. Give diagnostic criteria, if applicable Ok 33 Data sources/ 8* For each variable of interest, give sources of data and details of methods of

34 http://bmjopen.bmj.com/ measurement assessment (measurement). Describe comparability of assessment methods if there is 35 36 more than one group Ok 37 Bias 9 Describe any efforts to address potential sources of bias 38 Potential sources of bias are included in the discussion 39 Study size 10 Explain how the study size was arrived at 40 All recorded cases of ICH caused by FNAIT from the international No 41 IntraCranial Haemorrhage (NOICH) registry during the period 2001-2010, were 42 included. From the NOICH registry recording 592 FNAIT cases, 66 pregnancies on October 2, 2021 by guest. Protected copyright. 43 from 57 women were initially identified with suspected ICH. In 23 cases ICH or 44 FNAIT diagnosis could not be confirmed. In total, 43 confirmed cases of ICH due 45 to FNAIT were recorded. 46 Quantitative 11 Explain how quantitative variables were handled in the analyses. 47 variables Means with 95% confidence intervals (CI) or median with range were calculated 48 for all continuous variables 49 If applicable, describe which groupings were chosen and why NA 50 Statistical methods 12 (a) Describe all statistical methods, including those used to control for confounding 51 Ok 52 (b) Describe any methods used to examine subgroups and interactions NA 53 c 54 ( ) Explain how missing data were addressed 55 If necessary data to confirm the FNAIT or ICH diagnosis was missing, the case 56 was excluded from the final study population 57 (d) If applicable, explain how loss to follow-up was addressed NA 58 (e) Describe any sensitivity analyses NA 59 Results 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml1 BMJ Open Page 46 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially 2 eligible, examined for eligibility, confirmed eligible, included in the study, completing 3 follow-up, and analysed Ok 4 5 (b) Give reasons for non-participation at each stage OK 6 (c) Consider use of a flow diagram Not considered to be necessary 7 Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and 8 information on exposures and potential confounders Ok 9 10 (b) Indicate number of participants with missing data for each variable of interest 11 Was not included in manuscript due to space considerations, but may be 12 provided upon request 13 (c) Summarise follow-up time (eg, average and total amount) NA 14 Outcome data 15* Report numbers of outcome events or summary measures over time NA 15 For peer review only 16 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and 17 their precision (eg, 95% confidence interval). Make clear which confounders were 18 adjusted for and why they were included 19 Ok. Confounder-adjusted estimates were n.a. 20 b 21 ( ) Report category boundaries when continuous variables were categorized NA 22 (c) If relevant, consider translating estimates of relative risk into absolute risk for a 23 meaningful time period NA 24 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity 25 26 analyses NA 27 Discussion 28 Key results 18 Summarise key results with reference to study objectives Ok 29 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or 30 31 imprecision. Discuss both direction and magnitude of any potential bias Ok 32 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, 33 multiplicity of analyses, results from similar studies, and other relevant evidence Ok

34 http://bmjopen.bmj.com/ Generalisability 21 Discuss the generalisability (external validity) of the study results Ok 35 36 Other information 37 Funding 22 Give the source of funding and the role of the funders for the present study and, if 38 applicable, for the original study on which the present article is based Ok 39 40 41 *Give information separately for exposed and unexposed groups. 42 on October 2, 2021 by guest. Protected copyright. 43 Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and 44 published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely 45 46 available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at 47 http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is 48 available at http://www.strobe-statement.org. 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml2 Page 47 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Flow diagram of the study population 34 http://bmjopen.bmj.com/ 35 112x90mm (300 x 300 DPI) 36 37 38 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 48 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 Estimated time period for onset of ICH is shown for 41 cases of ICH caused by FNAIT. An arrowhead to the left indicates that the earliest time of onset cannot be estimated, only that the bleeding occurred before the 48 gestational age indicated on the x-axis. 49 90x132mm (300 x 300 DPI) 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 49 of 49 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on October 2, 2021 by guest. Protected copyright. 43 44 45 46 47 a) CT scan shows a very large intraparenchymal hemorrhage with mass effect and occupying most of the right frontal lobe in a term baby. Note a second but smaller hemorrhage on the left side adjacent to the 48 ventricle but separate from this. The attenuation of both hemorrhages indicates that the hemorrhage is 49 several weeks old at the time of imaging. We estimated the right hemorrhage to be 4 - 6 weeks old and the 50 left still a couple of weeks older. The pregnancy was considered normal, and no IVIG treatment was given. A 51 live boy was born vaginally at 40 gestational weeks with petecchiae. The platelet count was 6x109/L. 52 b) This CT shows marked ventriculomegaly, partly due to hydrocephalus, partly due to loss of brain tissue in 53 the left hemisphere. Note the very large ventricle on the left side occupying most of the left hemicranium. 54 The shape of the left lateral ventricle is irregular and that of residual from a previous intraventricular 55 hemorrhage with a paraventricular atrophic defect caused by an old periventricular hemorrhagic infarction. This pattern is pathognomonic for this condition even though no actual blood can be detected. The 56 interpretation of this image is that of an intraventricular hemorrhage associated with a periventricular 57 hemorrhagic infarction timed at 28 – 30 gestational weeks or earlier. Hydrocephalus was diagnosed 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 50 of 49 BMJ Open: first published as 10.1136/bmjopen-2012-002490 on 22 March 2013. Downloaded from 1 2 3 intrapartum because of breech presentation. A live boy was born at term with a platelet count of 4x109/L. 4 The child has severe cerebral palsy and a hydrocephalus shunt. 5 c) The CT scan shows multiple focal intraparenchymal hemorrhages throughout both cerebral hemispheres. 6 Most hemorrhages are quite small. Note also the extensive extracranial subgaleal hemorrhage overlying the 7 right hemicranium. All bleedings are of same age and maximum seven days old. It was noted that the fetus 8 was small on US scan at 25 weeks, but otherwise the pregnancy was considered normal. At 42 weeks a live 9 boy was born vaginally, with multiple petecchiaes and multiple retinal bleedings. The platelet count at delivery was 14x109/L, with nadir value 8x109/L. He received platelet transfusions. At time of discharge the 10 boy was described as alive and well, but we do not have data on long-term clinical outcome. 11 12 13 90x127mm (300 x 300 DPI) 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33