Germinal matrix hemorrhage–intraventricular hemorrhage (GMH-IVH) is a significant cause of neonatal brain injury, particularly in preterm infants, and is a major concern in neonatal radiology [1]. The incidence of GMH-IVH is inversely related to both gestational age and birth weight [2,3]. The germinal matrix is particularly vulnerable due to its fragile basal lamina, immature vasculature, and low levels of glial fibrillary acidic protein expression. This vulnerability is exacerbated by structural anomalies in the subependymal vein, hypoxia, and a lack of adaptability, all of which increase the risk of vascular rupture [4]. Such conditions often lead to hemorrhage in the small vessels of the germinal matrix, which can disrupt the ependymal lining and extend into the lateral ventricle [3,4]. GMH-IVH is typically classified into grades I-IV based on the severity, according to the classification system developed by Papile et al. [3].

The germinal matrix, which is located below the floor of the lateral ventricles, undergoes extensive cellular proliferation during early fetal life. It reaches its peak volume between 20 and 26 weeks of gestation before gradually regressing in a site-specific manner [5,6]. This reduction in volume is especially pronounced in the dorsal portion, whereas the ventral portion remains within the anterior caudothalamic groove until it fully involutes by 34 to 36 weeks of gestation [5-7].

Given the dynamic nature of the germinal matrix throughout gestation, it is understood that GMH presentations may vary between extremely preterm and late preterm infants, reflecting developmental differences [7]. In extremely preterm infants, hemorrhage can occur in the remaining posterior germinal matrix, termed posterior subependymal GMH (PS-GMH), located behind the typical grade I GMH-IVH found in the caudothalamic groove [7,8]. However, large PS-GMH has sometimes been misidentified as high-grade GMH-IVH, choroid plexus hemorrhage, or parenchymal injury due to an incomplete understanding of the germinal matrix [7].

Previous studies have demonstrated that abnormalities in the germinal matrix, identified via fetal imaging in the second trimester—including abnormal persistence, enlargement, and cavitation—are associated with significantly reduced fetal head size, underdevelopment of sulcation, and malformations in cortical development [9-11]. Additionally, experimental research has shown that germinal matrix hemorrhage in preterm infants (aged 18-34 weeks post-menstrual) correlates with decreased Ki67 immunoreactivity, suggesting a reduction in germinal cell proliferation [5]. This reduction may lead to impaired oligodendrocyte development near the germinal matrix, negatively impacting cerebral myelination. It could also disrupt the production of GABAergic interneurons for the neocortex and interfere with the generation of thalamic neurons, potentially resulting in adverse delayed effects [12,13]. However, the neurodevelopmental outcomes for preterm infants with PS-GMH identified through postnatal imaging remain uncertain.

This study aims to explore the effect of both the presence and size of PS-GMH, considered a mild form of hemorrhage, on the neurodevelopmental outcomes of extremely preterm infants.

The study included 221 infants, of which 192 (86.9%) were diagnosed with PS-GMH. This group consisted of 166 (86.4%) cases of small PS-GMH and 26 (13.5%) cases of large PS-GMH. Among these, 84 (43.8%) had bilateral PS-GMH, while 108 (56.3%) had unilateral PS-GMH. On the initial cUS, 21 of the 192 infants with PS-GMH (10.9%) also presented with typical anterior GMH (Fig. 3). Additionally, 76 infants (39.6%) did not exhibit typical anterior GMH initially but developed it on delayed follow-up cUS. The median interval between the initial and follow-up cUS was 36 days (interquartile range, 29 to 60 days). There were no instances of delayed PS-GMH at follow-up cUS. Neurodevelopmental outcomes were evaluated in 135 infants at corrected ages of 18-24 months. Comparisons between infants with and without neurodevelopmental outcomes revealed no significant differences in neonatal and maternal factors, or in cUS findings (Table 1).

The median timing for cUS examinations was 3 days (interquartile range, 2 to 5 days) after the initial assessment and 84 days (interquartile range, 74 to 93 days) for the term-equivalent assessment.

Clinical characteristics, cUS findings, and neurodevelopmental outcomes based on the presence and size of PS-GMH are presented in Table 2. Infants with PS-GMH were born at a significantly younger gestational age than those without PS-GMH (P<0.001). When comparing the groups with no PS-GMH, small PS-GMH, and large PS-GMH, there were significant differences in gestational age (P<0.001). However, post-hoc tests indicated no significant difference in gestational age between the small and large PS-GMH groups (P>0.99) (Fig. 4). Maternal chorioamnionitis occurred more frequently in infants with PS-GMH (P=0.016); however, this difference was not statistically significant when analyzed by the size of PS-GMH (P=0.053). No significant differences were observed in other neonatal and maternal factors with respect to the presence and size of PS-GMH.

During both initial and TEA cUS assessments, no significant differences were observed in the presence of grade II IVH, white matter hyper-echogenicity, ventriculomegaly, and periventricular leukomalacia between infants with and without PS-GMH. However, when comparing groups with no PS-GMH, small PS-GMH, and large PS-GMH, statistically significant differences were noted in the incidence of grade II IVH at the initial cUS (P=0.003) and ventriculomegaly at TEA cUS (P=0.026). Specifically, the large PS-GMH group exhibited a higher prevalence of grade II IVH during the initial cUS than the small PS-GMH group (P=0.006) (Figs. 5, 6). Nevertheless, there were no statistically significant differences in ventriculomegaly at TEA cUS based on PS-GMH status (Figs. 5, 6).

At corrected ages of 18-24 months, 47 infants (34.8%) were found to have at least one neurodevelopmental impairment. There was no significant difference in the proportion of neurodevelopmental impairments across different PS-GMH statuses (no PS-GMH, all PS-GMH, small PS-GMH, large PS-GMH), with rates of 31.3% (5/16), 35.3% (42/119), 33.0% (34/103), and 50.5% (8/16), respectively.

In comparison to the no PS-GMH group, the adjusted ORs for any neurodevelopmental impairment were 1.70 (95% CI, 0.40 to 7.26; P=0.471) in all infants with PS-GMH, 1.61 (95% CI, 0.37 to 6.93; P=0.526) in the small PS-GMH group, and 3.84 (95% CI, 0.62 to 24.00; P=0.150) in the large PS-GMH group (Table 3).

The inter-observer agreement for cUS findings in 221 patients showed substantial to almost perfect reliability (κ=0.67-0.97), with specific follows as possible: PS-GMH presence (κ=0.92), PS-GMH size (κ=0.94), grade II IVH (κ=0.97), white matter hyper-echogenicity (κ=0.67), ventriculomegaly (κ=0.77), and periventricular leukomalacia (κ=0.87).

The present study revealed that 87% of extremely preterm infants exhibited PS-GMH, with 86% having small PS-GMH and 14% having large PS-GMH. Additionally, 44% of these infants had bilateral PS-GMH, while 56% had unilateral PS-GMH. Infants with PS-GMH were born at earlier gestational ages and had a higher prevalence of maternal chorioamnionitis compared to those without PS-GMH. Significant differences were observed in the incidence of grade II IVH and ventriculomegaly among groups with no PS-GMH, small PS-GMH, and large PS-GMH. Specifically, the large PS-GMH group showed a higher prevalence of grade II IVH during initial cUS compared to the small PS-GMH group. This study is the first to assess neurodevelopmental outcomes in extremely preterm infants with PS-GMH. Regardless of size, PS-GMH did not predict adverse neurodevelopmental outcomes based on BSID and assessment of cerebral palsy at corrected ages of 18-24 months.

Despite advancements in neonatal care, the prevalence of GMH-IVH among preterm infants has remained unchanged, partly due to the higher survival rates of extremely preterm infants [23]. There is a significant correlation between the severity of GMH-IVH and earlier gestational ages, which frequently results in major neurological impairments such as cerebral palsy and post-hemorrhagic hydrocephalus [1,23,24]. Imaging studies conducted after GMH-IVH have shown considerable abnormal development in brain volume [25]. However, early detection and neuroprotective interventions can improve outcomes for these infants [26].

The GMH-IVH classification, which is based on the grading system developed by Papile et al. [3] over four decades ago, continues to be used primarily for the more mature preterm population. According to this system [3], grade I GMH-IVH is typically defined as a hemorrhage located in the subependymal region, particularly at the caudothalamic grooves. GMH-IVH may arise from any part of the residual germinal matrix, which has been found to extend to new, previously unrecognized locations in infants born as early as 28 weeks' gestational age. Recent studies by Snyder et al. [7] have shown that the survival of extremely preterm infants has led to the identification of a "posterior GMH," which reflects a fetal pattern of germinal matrix distribution not previously documented.

The germinal matrix, a transient embryonic structure in the fetal brain, is situated along the lateral wall of the frontal horns of the lateral ventricles and begins to form by the fifth week of gestation at the base of the telencephalon [27]. It is essential for the development of various brain structures, including the basal ganglia, thalamic and olfactory interneurons, and cortical GABAergic interneurons [27,28]. As gestation advances, the size of the germinal matrix shows an inverse correlation with the overall brain and cortical volumes, reaching its peak between weeks 20 and 26, and subsequently diminishing in a pattern that progresses from the dorsal to the ventral parts, completing its regression by weeks 34 to 36 [5,6]. In late preterm infants, GMH typically occurs in the ventral portion of the germinal matrix, near the caudothalamic groove. In contrast, in extremely preterm infants, particularly those born before 28 weeks of gestation, a significant portion of the germinal matrix persists in the dorsal area, where hemorrhage may present as hyper-echogenicity behind the caudothalamic groove [7,8]. This study highlights the importance of recognizing a new pattern of GMH-IVH in extremely preterm infants to avoid misdiagnosing other types of intracranial hemorrhage (Figs. 3, 5, 6), which could lead to unnecessary and invasive treatments. When the ventricles are distended with cerebrospinal fluid (CSF), observing a CSF cleft on parasagittal US images can help accurately differentiate PS-GMH from periventricular white matter, facilitating correct diagnosis and treatment planning [7,29] (Figs. 3, 5, 6).

Contrary to the present findings, Indrakanti et al. [29] observed linear or mildly lobulated hyper-echogenicity at the floor of the frontal horns of the lateral ventricle, which they interpreted as normal variations of the residual germinal matrix rather than signs of disease. However, their study did not establish a correlation between radiology and pathology. In contrast, Contro et al. [11] described the germinal matrix in fetal US as an ovoid-shaped area with iso-echogenicity similar to the surrounding white matter, challenging the interpretation of hyper-echogenicity as a normal variation. This discrepancy suggests that linear or nodular hyper-echogenicity observed along the subependymal layer on US could indeed indicate hemorrhages within the fragile residual germinal matrix, rather than merely normal variations.

This retrospective study is constrained by potential selection bias and a relatively small sample size, comprising only 221 extremely preterm infants from a single center. A significant limitation is the absence of pathology or magnetic resonance imaging (MRI) correlation. At the authors’ institution, brain MRIs are not routinely performed shortly after birth but may be conducted at TEA for specific cases. Detecting small PS-GMH can be challenging when the CSF space in the lateral ventricle is obliterated. Furthermore, differentiating between large PS-GMH and intraventricular blood clots that adhere to the subependymal lining can present a diagnostic challenge, although follow-up imaging typically assists in this differentiation.

In extremely preterm infants, the germinal matrix is larger than in more mature preterm infants, which may lead to more voluminous GMH. The Papile classification categorizes subependymal hemorrhages as grade I, regardless of their size. This classification becomes problematic when trying to differentiate between grade I and grade III hemorrhages, especially in cases of large PS-GMH. Due to their larger size and earlier gestational age, these hemorrhages do not align well with the severity spectrum outlined by the Papile classification. Thus, Snyder et al. [7] suggested that describing these hemorrhages as PS-GMH, including details about their location and size, offers a more accurate classification than simply categorizing them as grade I.

Although the findings indicate that PS-GMH does not seem to negatively impact neurodevelopmental outcomes at corrected ages of 18-24 months, there is a noticeable trend toward higher neurodevelopmental impairment in cases with large PS-GMH. This observation underscores the need for further research into the potential effects at toddler or early school age. Specifically, the influence of PS-GMH on subtle disabilities, such as late-onset high-prevalence/low-severity disabilities like attention-deficit hyperactivity disorder, specific neuropsychological deficits, and behavioral issues, needs further clarification. Additionally, it is crucial to recognize the weak correlation between BSID scores at corrected ages of 18-24 months and later school year outcomes [30,31]. Despite these limitations, this study offers valuable insights into the prognostic implications of PS-GMH detected via cUS in extremely preterm infants, highlighting the necessity for more comprehensive, long-term studies.

This study on extremely preterm infants reveals that PS-GMH, despite its high prevalence, does not predict neurodevelopmental outcomes at corrected ages of 18-24 months. The findings indicate that the size of PS-GMH, whether small or large, does not negatively impact the neurodevelopmental trajectory. This information is vital for radiologists and clinicians, emphasizing the importance of distinguishing PS-GMH from other types of intracranial hemorrhages to guide clinical decisions and effectively counsel families. Although associated with factors such as younger gestational age and maternal chorioamnionitis, the identification of PS-GMH should provide reassurance to healthcare providers and families regarding its lack of correlation with adverse neurodevelopmental outcomes within the study’s timeframe. This highlights the need for further research to investigate the long-term effects of PS-GMH beyond early childhood.