Subpial Hemorrhage of the Neonate

Abstract

BACKGROUND AND PURPOSE:

Subpial hemorrhage of the neonate (SHN) is a rare stroke subtype reported in few case series. Birth trauma and coagulopathy are commonly proposed etiologies. We evaluated our SHN patient cohort to expand current understanding

METHODS:

Cases of SHN were identified by keyword searches of the institutional database. The medical records and MRIs were reviewed.

RESULTS:

Seventeen cases were identified. Assisted delivery occurred in 12% of cases, and acute coagulation abnormalities occurred in 77%. SHN was located in the temporal lobe in 82%, with cytotoxic edema and medullary vein congestion and/or thrombosis subjacent to the hemorrhages in 100% and 76% of cases, respectively. Neurologic disability was present in 44% of survivors. Three patients had chronic coagulation abnormalities.

CONCLUSION:

In our cohort, clinical findings supporting a potential relationship with birth trauma were infrequent. The imaging findings suggest a non-arterial, deep venous pattern of hemorrhagic ischemia.

Introduction

Hemorrhagic stroke involving the subpial compartment is a rarely described entity. It was initially reported in 1972, described as bleeding between pia and displaced brain¹. In 2004, Huang and Robertson described seven cases of subpial and superficial parenchymal hemorrhage, confirmed by MRI². Other studies have included these unique hemorrhages among reports of more broadly defined neonatal hemorrhagic stroke (NHS)^3–6.

Proposed etiologies in subpial hemorrhage of the neonate (SHN) include birth trauma and coagulopathy. In this study, we evaluated potential perinatal risk factors, imaging, and outcomes to better characterize this rarely studied hemorrhagic stroke.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Ascertainment of Cases

This retrospective study was approved by the Colorado Multiple IRB. Clinical and radiographic data was de-identified. Consent was not obtained from included patients. Potential cases were identified through keyword searches of the hospital database using mPower™ Clinical Analytics with the terms “subpial”, “subarachnoid”, and “hemorrhage”. Searches were limited to brain MRIs performed in neonates (birth to 28 days) between 2006–2016, resulting in a total of 382 studies. Axial T2 and GRE/SWI sequences for each scan were reviewed independently by two pediatric neuroradiologists to confirm presence of subpial hemorrhage defined as 1) hemorrhage closely opposing the underlying sulci 2) pooling of blood products relatively localized rather than spreading along the convexity, as is typical in subarachnoid hemorrhage (figure 1). When available, MRA and MRV images were assessed. Two neonatal/fetal neurologists reviewed the electronic medical records for prenatal and postnatal clinical data.

Figure 1.

A) Axial T2-weighted image demonstrates subpial hemorrhage (*) closely opposing the underlying sulci. B) Axial GRE image reveals prominence of the medullary veins within the subjacent parenchyma (→). C) Axial DWI image illustrates cytotoxic edema throughout the involved parenchyma (>).

Results

Clinical Data

Our final cohort included 17 cases of SHN. Each of the following pregnancy complications occurred once (6%) in our cohort: chorioamnionitis, oligohydramnios, premature preterm rupture of membranes plus HELLP syndrome, pre-eclampsia, and intrauterine growth restriction. Perinatal course and outcomes are shown in table 1. The majority (82%) of deliveries were vaginal, with only two deliveries having either vacuum or forceps assistance. Signs of fetal distress (fetal heart rate abnormalities or meconium) occurred in 8 patients (47%).

TABLE 1:

Perinatal Course and Outcomes

Labor Events	n(%)
Vaginal Delivery	14(82%)
Emergent C-Section	3(18%)
Assisted Deliveryϕ	2(12%)
Evidence of Fetal Distress	8(47%)
Abnormal fetal monitoring	6
Meconium	5
Birth
Gestational Age, mean ± SD	38.6 weeks ± 2.12)
Male, n(%)	11(65%)
Birth Weight, mean ± SD	3.2 ± 0.7 kg
Apgar score @ 1 minute, median (IQR)	7 (3–8)
Apgar score @ 5 minutes, median (IQR)	8 (5–9)
Delivery Room Resuscitation†	8 (47%)
Clinical Presentation	n(%)
Clinical seizure	7(41%)
Apnea	7(41%)
Clinical HIE	2(12%)
Postnatal Complications (first 7 postnatal days)	n(%)
Electrographic seizures	6(35%)
Treatment for Sepsis	3(18%)
Required Intubation	9(53%)
Need for neurosurgical Intervention	2(12%)
Outcome
Died, n(%)	1(6%)
Length of NICU stay in days, mean ± SD	20.5 ± 15.6
Neurologic exam at discharge abnormal, n(%)	5(29%)
Age at most recent follow up in months, median(IQR)	36(20–60)
Abnormality at most recent follow up (out of 16 surviving)	7(44%)
Motor	3
Language delay	4
Other	1
Remote symptomatic epilepsy	1(6%)
Coagulation Abnormalities	n(%)
Acute Findings (first 7 postnatal days):
Thrombocytopenia (<150×103/μL)	6(35%)
Elevated PT (>16.5 seconds) or PTT (>48 seconds)	4(24%)
Elevated D-dimer (>0.5μg/mLFEU)	9(53%)
Low Fibrinogen (<150mg/DL)	5(29%)
Chronic Persistent Findings
Hypercoagulable disorder	3(18%)‡
Thrombophilia	1(6%)§

^ϕOne vacuum, one forceps.

^†Additional resuscitation beyond drying, stimulation, blow-by O₂.

^‡one patient had ATIII deficiency, one patient had persistently elevated anti-Cardiolipin IgM at 14-months of age, one patient had transiently elevated anti beta Gp1 antibodies at 19-months.

^§Factor VII deficiency

The most common presenting signs were apneic events and clinical seizures (41% each). Need for intubation and electrographic seizures were the most common acute complications occurring within the first week of life (53% and 35% respectively). Acute coagulation abnormalities were common (77%) within the first 7 postnatal days, the most frequent being elevated D-dimer (53%) and low fibrinogen (29%). Acute thrombocytopenia was present in 35% of patients. Only one patient was later diagnosed with a hypercoagulable disorder (Anti-Thrombin III Deficiency), and one patient was diagnosed with thrombophilia (Factor VII deficiency). Two additional patients had persistent findings of unclear significance: one with elevated anti-cardiolipin IgM, and one with elevated beta glycoprotein1 antibodies (normalized by 21 months of age).

One patient from our cohort died during the initial hospitalization due to complications of congenital heart disease. Discharge neurologic exam was abnormal in 5 patients (29%), consisting of mildly low tone in 4 patients and visual hemi-neglect in one patient. None of the patients required gastrostomy tube at discharge. Follow up information was available for the 16 surviving patients (median age 36 months, IQR 20–60 months), seven of which (44%) had neurologic deficits at follow up. Three patients (19%) had mild hemiparesis. Four patients (25%) had delays in either expressive or receptive language, and one patient was diagnosed with mild intellectual disability. Only one patient developed remote symptomatic epilepsy. There were no patients who had recurrent hemorrhage at follow up.

Imaging Findings

Results of MRIs are provided in table 2. The temporal lobe was involved in 82% of cases. Intraparenchymal cytotoxic edema was subjacent to the subpial hemorrhages in all cases. Concomitant hemorrhage in other compartments was common, with intraparenchymal and intraventricular most common in 82% and 65% respectively. Prominence of the medullary veins was seen in 76%. The term “prominent” was used given the difficulty of distinguishing between medullary vein engorgement and thrombosis. A fan-shaped appearance of restricted diffusion or hemorrhage termed the “iris sign” - thought to be a biomarker of medullary venous thrombosis - was seen in 65% of patients⁷. There were no MRA abnormalities (n=13 performed). MRV revealed a thrombosed transverse venous sinus in one patient (n=10 performed).

TABLE 2:

MRI analysis of cases of neonates with subpial hemorrhage

Initial Imaging
Hemorrhage Volume, mean* Concomitant Hemorrhage	20965 mm3 (SD 22025 mm3)
Intraparenchymal	14 (82%))
Intraventricular	11 (65%)
Epidural	0
Subdural	7 (41%)
Lobe Involved	82% temporal, 29% occipital, 18% frontal, 12% parietal
Adjacent Suture	8 (47%)
Soft Tissue Swelling	2 (12%)
Cytotoxic Edema	17 (100%)
Prominent Medullary Veins	13 (76%)
MR Angiography	n=13, all negative
MR Venography	n=10, one positive†
Iris Sign	65% (n=11)

^*L x W x H x π/6

^†thrombosed transverse sinus

Discussion

SHN is unique to the neonatal time period, and is pathologically distinct from subarachnoid hemorrhage¹. SHN represents a small subset of NHS, and likely differs in mechanism from other NHS with unique risk factors.

Peripartum Factors

Although subdural or subarachnoid hemorrhages are reported with traumatic or assisted delivery, associations with parenchymal hemorrhage or IVH are mixed^{2, 5, 8}. Prior studies of all-compartment NHS demonstrated that proxies for traumatic birth, such as instrument assisted birth, were not associated with higher odds of NHS ³. In this study, fetal distress was seen in nearly half of our cases, but assisted delivery was rare. This suggests that fetal distress preceding delivery may be a symptom of in-utero SHN, rather than SHN resulting from traumatic delivery.

Hematologic Abnormalities

Acute thrombocytopenia and coagulation abnormalities were common in our cohort, similar to other cohorts of NHS, particularly those with hemorrhagic transformation of venous ischemia ⁴. Ultimate diagnosis of hypercoagulable disorder or hemophilia was rare in our cohort, as was recurrence or extension of hemorrhage. These data suggest that SHN is an isolated in-utero event with transient risk factors.

Outcomes

Only one patient died, due to factors related to congenital heart disease. As such, we theorize that neonatal death resulting from subpial hemorrhage itself is uncommon. Outcomes after NHS vary widely depending on the etiology.⁶ While 44% of surviving patients in this study had neurologic deficits at follow up, these tended to be relatively mild and only one patient developed remote symptomatic epilepsy.

Imaging

Our cases of SHN did not follow a typical arterial territory and all MRAs performed were negative. Bleeds occurred in known distributions of medullary venous territories. In most of our cases, the medullary veins were noted to be enlarged, denoting congestion and/or thrombosis. We also found that both intraventricular and intraparenchymal hemorrhages were frequent in our cohort, both of which are common hemorrhage locations in neonates with venous thrombosis^{9, 10}. We therefore hypothesize that medullary venous congestion may account for many cases of SHN.

Study Limitations

Our study was limited by small numbers and lack of control group.

Conclusion

In a small institutional cohort, fetal distress rather than birth trauma was common, suggesting SHN as a pre-delivery rather than a during- or post-delivery occurrence. The presence of subpial hemorrhage with underlying parenchymal ischemia and/or hemorrhage in a non-arterial, deep venous distribution suggest there may be a larger role for venous ischemia in the pathogenesis of SHN.