Outcome of antenatally diagnosed intracranial hemorrhage: case series and review of the literature




Prenatal diagnosis of intracranial hemorrhage (ICH) has been widely reported. Hemorrhages may occur either within the cerebral ventricles, subdural space or infratentorial fossa. The aim of this study was to determine the sonographic criteria for the diagnosis of fetal ICH, the role of in utero magnetic resonance imaging (MRI) and the outcome of this condition.


The archives of our ultrasound laboratory and the literature were searched for all cases of antenatally diagnosed ICH. A grading system was used to classify the intraventricular lesions as suggested in postnatal sonographic studies.


Adding our series of 16 fetuses to the 93 cases identified in the literature, a group of 109 fetal ICHs was obtained: 89 were intracerebral (79 intraventricular and 10 infratentorial) and 20 were subdural hemorrhages. Intraventricular lesions were mostly classified as severe (32 each for Grades III and IV). In 27 cases antenatal MRI was performed additionally to ultrasound and confirmed the sonographic findings. Of the entire group, 65 infants (59%) were reported to be alive 1 month after birth (51 intraventricular hemorrhages, three infratentorial hemorrhages, 11 subdural hematomas). At 12 months, of the 48 infants whose follow-up was available, 25 or 52% were judged neurologically normal (17/36 or 47% among the intraventricular hemorrhages, 6/9 or 66% among the hematomas, and 2/3 or 66% among the infratentorial hemorrhages).


Fetal ICH may be accurately identified and categorized by antenatal sonography. The outcome is usually poor, especially for those fetuses affected by higher-grade intraventricular hemorrhages. Copyright © 2003 ISUOG. Published by John Wiley & Sons, Ltd.


Intracranial hemorrhage (ICH) is a common event in premature infants1, 2. It may occur in utero also, and prenatal diagnosis by either sonography or magnetic resonance imaging (MRI) has been reported3–64. Factors predisposing to in utero ICH include a variety of conditions, mostly maternal trauma and fetal coagulation disorders55. In many cases, however, the cause is not identified. The incidence is unclear, although an estimate of 1 in 10 000 pregnancies has been suggested63.

Sonographic findings of fetal ICH are variable and at times subtle. In general, this condition is considered difficult both to identify and to differentiate from other intracranial lesions. However, with continuous improvement of ultrasound equipment, the number of fetal diagnoses is likely to increase.

The aim of our study was to determine from our own series of 16 fetuses and from a review of the available literature the sonographic criteria for the diagnosis of fetal ICH, the role of MRI, the clinical implications and outcome of this condition.


The archives of our ultrasound laboratory were searched retrospectively from 1981 for all cases of antenatally diagnosed ICH. No attempt was made to look for false negatives.

Ultrasound examinations were always performed transabdominally using a multiplanar approach as previously described65, with 3.5–7.5-MHz probes. Since 1987, when permitted by a low position of the fetal head, neurosonography was performed transvaginally using 5–10-MHz probes. From 1999, all cases with suspected fetal ICH were offered MRI. ICHs were subdivided into intracerebral and extracerebral (subdural hematomas). Intracerebral hemorrhages were further subdivided into intraventricular and infratentorial (occurring within the posterior fossa).

Intraventricular hemorrhages were categorized following the classification commonly used in neonates1 as follows:

  • Grade I: limited to subependymal matrix;

  • Grade II: clear spill-over to ventricles, but filling less than 50% of the lateral ventricle and without ventriculomegaly;

  • Grade III: spill-over to the ventricle, with flooding of 50% or more of one or both lateral ventricles and ventriculomegaly; and

  • Grade IV: Grades I, II or III with hemorrhage in a large part of the periventricular parenchyma.

In our own series we decided to classify intraventricular hemorrhages associated with a transverse diameter of the lateral ventricular atrium of < 15 mm as Grade II, and those with a diameter > 15 mm as Grade III.

The clinical data of the pregnancies, in addition to a detailed follow-up of all cases, were obtained in all instances through medical records and/or telephonic interviews with the physicians and/or parents of the affected infants.

A literature search was performed through MEDLINE using the following keywords: intracranial hemorrhage; intracerebral hemorrhage, subdural hematoma, infratentorial or cerebellar hemorrhage and fetus; prenatal diagnosis. Additional sources were identified through cross-referencing.

Only reports where the diagnosis of ICH had been achieved antenatally and confirmed at follow-up were considered. When available, pregnancy outcome and neurological follow-up of the reported cases were derived. When indicated by the authors, maternal or fetal predisposing conditions were considered.


In our center, 16 cases of fetal ICHs including 14 intracerebral hemorrhages and two subdural hematomas were detected antenatally from 1981 to 2001 (Table 1). Two of these cases had been previously reported66.

Table 1. Details of cases comprising present series
CaseWeeksType of hemorrhagePossible etiologyOutcome
  • *

    Previously reported case66. DQ, developmental quotient; IVH, intraventricular hemorrhage; Rh, rhesus.

132IVH Grade IINoneMild ventriculomegaly, normal at birth, lost at follow-up
222IVH Grade IINoneNormal neonatal neurosonogram
321IVH Grade IINoneNormal neonatal neurosonogram
424IVH Grade IIHydropsOvert ventriculomegaly at birth, ventriculoperitoneal shunting
529IVH Grade IIINonePregnancy termination, pathology confirmation
628IVH Grade IIINoneIntrauterine death, pathology not available
729IVH Grade IIICar accident at 25 weeksLive birth, porencephalic cyst, ventriculoperitoneal shunting, DQ < 50 at 4 years
833IVH Grade IIINoneLive birth, ventriculoperitoneal shunting, neurologically normal at 2 months
924IVH Grade IVIn utero death of the co-twin in a twin-to-twin transfusionLive birth, ventriculomegaly, neurologically normal at 2 months
1032IVH Grade IVAlloimmune thrombocytopeniaLive birth, porencephalic cyst, ventriculoperitoneal shunting, DQ < 50 at 4 years*
1129IVH Grade IVNoneLive birth, porencephalic cyst, ventriculoperitoneal shunting, DQ < 50 at 2 years*
1222IVH Grade IVNonePregnancy termination, pathology confirmation
1322InfratentorialHydrops, anemia, Rh alloimmunizationPregnancy termination, pathology confirmation
1422InfratentorialHydrops, anemia, Rh alloimmunizationLive birth, cerebellar hypoplasia, neurologically normal at 12 months
1523Subdural hematomaMaternal traumaLive birth, surgical drainage, neurologically normal at 10 months
1624Subdural hematomaIn utero death of the co-twin, possible twin-to-twin transfusionPregnancy termination, pathology confirmation

Intracerebral hemorrhages included 12 intraventricular and two infratentorial hemorrhages. Intraventricular hemorrhages included four cases each of Grade II, Grade III and Grade IV.

Intraventricular hemorrhages were diagnosed by the demonstration of findings suggestive of blood clots within the ventricles (Figure 1). Similarly to neonatal sonographic studies, fresh blood clots appeared as homogeneously echogenic areas, separate from the choroid plexus. In the following weeks blood clots developed a complex appearance, with an external echogenic lining and an internal sonolucent core (Figure 1). Of four fetuses with a Grade II intraventricular hemorrhage, three had borderline enlargement of the ventricles. In two of these cases the images interpreted as clots disappeared and ventriculomegaly resolved throughout gestation. The infants had negative neurosonograms at birth and appeared clinically normal at a short-term follow-up. In one case, associated with non-immune hydrops, the ventricles were normal at the first scan but progression of ventriculomegaly was noted and after birth the infant underwent a ventriculoperitoneal shunt.

Figure 1.

Sonographic demonstration of intraventricular hemorrhage. (a) Coronal and (b) sagittal views in a mid-trimester fetus (21 weeks) with a recent Grade II hemorrhage; the blood clot (arrows) appears as a uniformly echogenic collection close to the choroid plexus; the lateral ventricle is mildly enlarged. (c) Coronal scan in a third-trimester fetus (33 weeks) with an old Grade III intraventricular hemorrhage; the blood clot has developed an anechoic core.

In four cases a Grade III hemorrhage was diagnosed following the demonstration of images suggestive of blood clots within very enlarged lateral ventricles, with seemingly normal cerebral parenchyma. In two of these fetuses, at the first observation the dilated lateral ventricle had an echogenic lining, and the contralateral ventricle was either normal in size or mildly enlarged (Figure 2). In the former case, serial examinations were performed demonstrating a progressive decrease in size of the clot, disappearance of the echogenic lining of the ventricle and progressive dilatation of the contralateral ventricle. In the latter case, severe hydrocephalus due to aqueductal obstruction was demonstrated after birth and the infant underwent shunting. Intrauterine MRI was performed in two cases within 7 days of sonographic diagnosis and this confirmed the ultrasound findings (Figure 3).

Figure 2.

Grade III intraventricular hemorrhage in a 34-week fetus in an axial view (a) and a posterior coronal view (b). The lateral ventricle containing the blood clot has a bright echogenic lining (arrows) presumably representing ventriculitis.

Figure 3.

Grade III intraventricular hemorrhage. (a,b) Intrauterine T2-weighted magnetic resonance images (MRI) demonstrating ventriculomegaly and a blood clot (arrows) within one of the lateral ventricles in a 34-week fetus. (c) Postnatal MRI in the same case demonstrating considerable regression of the blood clot (arrow) and worsening of the ventriculomegaly; aqueductal stenosis was present and the infant underwent ventriculoperitoneal shunting soon after birth.

In four cases a Grade IV hemorrhage was suspected following the demonstration of diffuse intracerebral echogenicity obscuring one or both lateral ventricles and projecting into the surrounding cerebral parenchyma. In one case, serial examinations were performed. Ten days after the first examination, severe ventriculomegaly, intraventricular images suggestive of blood clots and irregular cerebral parenchyma surrounding the ventricles were seen. The diagnosis was confirmed by in utero MRI and by pathology after termination of pregnancy. In the remaining three cases a Grade IV hemorrhage was diagnosed because of the demonstration of a porencephalic cyst containing an image suggestive of a blood clot (Figure 4).

Figure 4.

Grade IV intraventricular hemorrhages. (a) Coronal and (b) sagittal views in a mid-trimester fetus (24 weeks) the day after the death of a monochorionic co-twin in the context of twin transfusion. There is a large echogenic collection involving one entire lateral ventricle and extending to the surrounding parenchyma. A porencephalic cyst and hypertensive hydrocephalus developed subsequently. (c) A large fetal porencephalic cyst in a 30-week pregnancy that was later found to be complicated by alloimmune thrombocytopenia. A blood clot is visible within the cyst.

Finally, infratentorial lesions were demonstrated in two mid-trimester fetuses. One pregnancy was electively terminated and pathology confirmed the presence of a hemorrhage within the cisterna magna and involving the cerebellum (Figure 5). In the latter case the couple opted to continue the pregnancy: progressive hypoplasia of the cerebellar hemisphere affected by the hemorrhage was demonstrated either by subsequent ultrasound scans that were serially performed throughout the pregnancy or by in utero MRI at 32 weeks of gestation (Figure 6). At birth, neuroimaging showed a hemicerebellar hypotrophy with no apparent clinical or neurological compromise of the infant who is now 1 year old.

Figure 5.

This mid-trimester fetus (22 weeks) was referred with hydrops and severe anemia (hemoglobin concentration 1.5 g/dL) due to severe Rh alloimmunization. The first intrauterine transfusion was successful. However, a few days later, ultrasound demonstrated the presence of an echogenic collection in the posterior fossa involving part of the cerebellum and the cisterna magna (arrows in a and b). The couple opted for pregnancy termination and pathology confirmed the presence of an infratentorial hemorrhage (arrow in c). 3v, third ventricle.

Figure 6.

Similarly to the case illustrated in Figure 5, this 22-week fetus was referred because of hydrops and severe anemia (hemoglobin concentration 1.8 g/dL) due to severe Rh alloimmunization. After the first successful intrauterine transfusion, an echogenic area was noted in one of the cerebellar hemispheres (arrows in a and b). The couple opted to continue the pregnancy. At 32 weeks in utero magnetic resonance imaging demonstrated progressive hypoplasia of the affected cerebellar hemisphere (arrow in c). This was confirmed after birth. 3v, third ventricle.

In two fetuses a subdural hematoma was diagnosed. In both cases ventriculomegaly of variable degree was associated with distension of the subarachnoid space that appeared densely corpusculated (Figure 7).

Figure 7.

This 24-week fetus was referred after the death of a monochorionic co-twin presumably affected by twin–twin transfusion syndrome. (a) Sonography revealed mild ventriculomegaly and distension of the subarachnoid space both at the level of the Sylvian fossa and interhemispheric fissure (arrows). The subarachnoid space also had a densely corpusculated appearance. (b) After termination of pregnancy, subdural hemorrhage was confirmed (arrows).

A plausible cause for the fetal ICH could be clearly identified in 7/16 cases (43%). One fetus with Grade IV hemorrhage was found after birth to have severe alloimmune thrombocytopenia. A Grade III hemorrhage was identified following a car accident involving the mother, and a subdural hematoma after a maternal trauma due to a domestic accident. A subdural hematoma as a Grade IV intraventricular hemorrhage occurred after the death of a co-twin with a monochorionic placentation. Both cases of infratentorial hemorrhage occurred in mid-trimester fetuses with severe anemia (1.5 and 1.8 g/dL) and hydrops as a consequence of severe rhesus (Rh) alloimmunization. Images suggestive of hemorrhage were documented a few days after the first successful intrauterine transfusion.

Reviewing the literature, 93 cases with antenatal ultrasonographic diagnosis of ICH were found, including 75 cases of intracerebral hemorrhage, 17 cases of subdural hematoma and one case of both3–64. Of the 75 cases of intracerebral hemorrhage the vast majority were classified as being of intraventricular origin (n = 67), the remaining cases having a pericerebellar or infratentorial location (n = 8).

By adding our own series, the total number of cases was 109. The mean gestational age at diagnosis was 30 (range, 20–37) weeks. In 27 fetuses MRI was performed in addition to ultrasound and confirmed the cranial findings. A maternal or fetal predisposing condition was identified in 48/109 cases (44%), and more precisely in 36 intracerebral and 12 subdural hematomas. A history of maternal trauma was present in 10 cases while a hemorrhagic diathesis was present in 19 fetuses (including nine with alloimmune thrombocytopenia). Only in two fetuses with antenatal diagnosis of subdural hematoma was an antecedent maternal trauma reported. The majority of ICHs were classified prenatally as intraventricular hemorrhages of either Grade III or IV (32 cases in each group).

Prenatal data and pregnancy outcome of the whole series are presented in Table 2. Intrauterine death occurred in 17 cases (15 intraventricular hemorrhages, of which 14 were classified as Grade III or greater, one combined intracerebral and subdural hemorrhage and one subdural hemorrhage).

Table 2. Prenatal data and pregnancy outcome (total)
 TotalPregnancy outcome
  • *

    Including one case of death intrapartum. IVH, intraventricular hemorrhage; IUD, intrauterine death; LB, live birth; TOP, termination of pregnancy.

Intracerebral 89 91565
 IVH I  6 0 1 5
 IVH II*  9 1 0 8
 IVH III 32 0 230
 IVH IV 32 312*17
 Infratentorial 10 5 0 5
Subdural hematoma 19 2 116
Both  1 0 1 0

Of the 81 fetuses (65 with intracerebral and 16 with extracerebral hemorrhage) that were liveborn, 16 died in the neonatal period. No losses were reported in fetuses with intracerebral hemorrhages of low grade (Grades I and II) whereas two deaths were registered in the subgroup with an infratentorial lesion.

Postnatal and neurological follow-up data were available for only 48/65 infants who survived the neonatal period (Table 3). At a mean age of 11.6 (range, 1–48 months) 25 infants were reported to be neurologically normal, while in 19 infants there was evidence of neurodevelopmental delay, either severe (n = 13) or mild (n = 6). Infantile death was described in four cases.

Table 3. Follow-up at 11.6 (range, 1–48) months of the available cases
 TotalNormalMild handicapSevere handicapDead
  1. IVH, intraventricular hemorrhage.

 IVH I 4 40 00
 IVH II 7 40 30
 IVH III18 85 50
 IVH IV 7 10 33
 Infratentorial 3 20 10
Subdural hematoma 9 61 11
Both 0 00 00

Two-thirds of the neonates (6/9) who had had an antenatal diagnosis of subdural hematomas were judged neurologically normal. In the 36 cases of intraventricular hemorrhage, outcome and neurological development at 13.3 months was found to depend upon the grade of the hemorrhagic lesion. All eleven infants whose intracranial hemorrhage had been prenatally staged as Grade I or II were alive, eight of them being neurologically normal (72%). Conversely, of the infants with Grade III and IV hemorrhage, 22/25 survived. Of these, 13 were handicapped and only nine (41%) were considered neurodevelopmentally normal. Finally, two of the three infants with a prenatal diagnosis of cerebellar hemorrhage were apparently normal, although their follow-up was short.


Our results suggest that antenatal ultrasound allows an accurate diagnosis of fetal ICH. The findings are variable and change with time. A new hemorrhage appears as a brightly echogenic collection without posterior shadowing. Over the following days blood clots develop a complex echogenic texture with an external echogenic lining and an internal sonolucent core. Intraventricular blood clots are usually associated with distension of the ventricles, which initially demonstrate a typical echogenic lining. Intraventricular hemorrhages may undergo spontaneous resolution or may obstruct the cerebrospinal fluid circulation, usually at the level of the aqueduct of Sylvius, resulting in ventriculomegaly. Involvement of the cortex can be predicted by demonstration of extension of the echogenic collection to the periventricular parenchyma in the early stages and formation of a porencephalic cyst usually about 2 weeks after the hemorrhagic event. A bright area around the cerebellum may prompt the diagnosis of an infratentorial hemorrhage.

In our hands ultrasound correlated well with the postnatal diagnosis, with the exception of only two cases in which resolution occurred in utero. Such evolution is, however, compatible with the natural history of these lesions and has been reported previously49.

The role of MRI in the evaluation of fetal cerebral anomalies is controversial67. Most of the cases from both review of the literature and our own experience were examined antenatally only with ultrasound. Our study therefore does not allow us to draw definitive conclusions on the relative role of ultrasound versus MRI. In our own experience the sonograms were always diagnostic, and MRI, when performed, did not add further information. Conversely, MRI always proved accurate. It also has the additional advantage that in T1-weighted sequences blood collections appear as areas of typically increased intensity68. It is clear that there is certainly a role for this technology, at least in those cases where antenatal sonography is inconclusive.

Our results and the review of the literature suggest that prenatally diagnosed ICHs have a poor outcome. About 40% of fetuses die either in utero or within the first month after birth. Among the survivors, less than half appear neurodevelopmentally normal at short-term follow up.

The survival rate is similar for subdural hematomas and intracerebral hemorrhage. At a follow-up of 1 year a slightly better outcome was observed with subdural hematomas than with intracerebral hemorrhage (normal development in 67% vs. 48% cases). Similarly to postnatal studies, the outcome of fetal intraventricular hemorrhage was strongly related to the grade of the lesion. Perinatal mortality with Grades I–II was 1/14 (7.1%) vs. 27/61 (44%) with Grades III–IV.

However, experience with Grade I and II hemorrhages was limited to a small number of cases, and was probably inadequate to provide firm prognostic figures. Indeed, complete regression of this lesion is possible. In two fetuses in our series and in a further case previously reported49 progressive disappearance of abnormal sonographic findings was observed, always in association with a normal postnatal outcome. Conversely, intrauterine progression of hemorrhage from a low to high grade is also possible as observed in our series and in previously reported cases11.

It is noteworthy that in three of our cases, unilateral borderline ventricles were identified in association with sonographic findings suggestive of an intraventricular hemorrhage. In two of these fetuses the signs of intraventricular hemorrhage disappeared during the course of the pregnancy, followed later on by restoration of normal ventricular size. We postulate that an intraventricular hemorrhage of low grade may be responsible for at least some cases of borderline ventriculomegaly, particularly the unilateral ones. The pathophysiological mechanism underlying transient ventricular dilatation is unclear. We speculate that this may be the consequence of obstruction to cerebrospinal fluid at the level of the foramina of Monro (in both of our cases blood clots were seen at the level of the bodies of the lateral ventricles). This may provide an explanation for the puzzling cases in which a transient fetal cerebral ventriculomegaly was identified69.

A predisposing condition to the in utero occurrence of ICH was identified or suspected in 47% of the cases, most frequently hematological disorder of the fetus and maternal trauma. An association with these risk factors has been noted either for intracerebral hemorrhage (12 coagulation disorders and eight trauma) or extracerebral hemorrhage (six coagulation disorders and two trauma). On the basis of these results, prenatally diagnosed subdural hematomas do not seem to be strictly related to a traumatic cause, as reported for those with perinatal onset48. Intrauterine hypoxia is commonly recognized as another possible cause of antenatal hemorrhage3, 63.

In four of the cases herein reported, the development of the hemorrhage was related to the presence of fetal anemia (two immune hydrops, two deaths of the co-twin with monochorionic placentation) as previously described22, 38, 54. The pathophysiology is unclear. We speculate that the hyperdynamic circulation that is well documented in severely anemic fetuses may have been the cause of the disruption of intracranial vessels. The documentation of a cerebral (in both cases infratentorial) hemorrhage in two severely anemic fetuses because of Rh alloimmunization is noteworthy. The occurrence of abnormal neurodevelopment in survivors after intrauterine transfusion has been previously documented70. The pathophysiology of brain damage in such cases is debated. We provide evidence that in some cases at least brain injury may occur prenatally. We believe that our observation suggests a role for fetal neurosonography in pregnancies with severe Rh alloimmunization undergoing intrauterine transfusion.

In some cases with Grade III hemorrhages, intraventricular clots disappeared during intrauterine life and only obstructive hydrocephalus was documented after birth. This observation may help to explain at least some cases in which severe hydrocephalus only appears in late gestation.

To the best of our knowledge this is the largest review on antenatally diagnosed ICHs. Most lesions occur sporadically and, apart from the cases in which a specific fetomaternal condition may be identified, the risk of recurrence is very low. According to the available data, the parents of affected fetuses must be counseled about a poor outcome following the diagnosis of either a Grade III or IV intraventricular hemorrhage or a subdural hemorrhage. Whether the unfavorable prognosis is entirely the consequence of the cerebral lesion or is influenced by the underlying disease causing the hemorrhage remains unclear. However, half of the fetuses will die within the first month of postnatal life and only half of these will be neurologically normal at short-term follow-up.

The outcome of fetuses with Grade I–II intraventricular hemorrhages or infratentorial hemorrhages is less clear given the limited number of cases reported thus far. As the sonographic findings in these cases are very subtle we also expect that many cases will escape antenatal detection. A correlation may exist between borderline ventriculomegaly, particularly when unilateral, and low-grade intraventricular hemorrhages.

The diagnostic work-up of a pregnancy complicated by sonographic evidence of an intracranial hemorrhage should include an interview focusing on drug use or recent trauma (motor vehicle accident, fall onto the abdomen, others) and laboratory tests for the possible presence of a fetal platelet disorder. This should include maternal platelet count and testing for both maternal alloimmunity and isoimmunity. Antenatal hemorrhage is possible with isoimmune thrombocytopenia but is exceedingly rare71. Conversely, it is frequently associated with the alloimmune type72. Treatment of alloimmune thrombocytopenia is still a matter of debate and the interested reader is referred to specific publications on this topic73–75. Whether treatment would be indicated in the presence of an established fetal intracranial lesion is uncertain. The optimal mode of delivery of an infant with sonographic evidence of an ICH is equally uncertain. Severe lesions usually have a poor prognosis and conservative management may be offered to the couples. There are no data to indicate that a Cesarean delivery may ameliorate the outcome of those infants with less severe lesions71.