Neurodevelopmental effect of intracranial hemorrhage observed in hypoxic ischemic brain injury in hypothermia-treated asphyxiated neonates - an MRI study

Hypoxic ischemic encephalopathy (HIE) secondary to perinatal asphyxia is the most common cause of neonatal encephalopathy (NE) [1]. In term or near-term infants (≥36 gestational weeks) with clinical signs of moderate to severe HIE, the current standard is therapeutic whole body hypothermia (33.5 ± 0.2 °C core temperature) for 72 h started within 6 h of life decreasing mortality and morbidity [2, 3], and improving neurologic outcome in survivors [4]. A meta-analysis of several major randomized trials including the National Institute of Child Health and Human Development (NICHD) study [3] and the Total Body Hypothermia (TOBY) trial [4] has demonstrated a reduction in death and neurological impairment at 18 months following hypothermia [5]. The results of the Infant Cooling Evaluation Collaboration (ICE) trial, a multicenter, international, randomized controlled trial shown the reduction of the risk of death and major sensorineural disability at 2 years of age and concluded that whole-body hypothermia is a safe and effective neuroprotective method with minimal adverse effects [6, 7]. However hypothermia treatment also has some consequences that the clinicians need to be aware of during treatment. A Cochrane Collaboration review on the safety of therapeutic hypothermia has shown thrombocytopenia and hypotension to be the major side effects of this therapeutic approach [8]. As cooling may result in decreased platelet counts, it may lead to prolongation of coagulation carrying a potentially higher risk for intracranial hemorrhage (ICH) in HIE newborns [9]. Besides, asphyxia very often leads to impaired cerebral autoregulation [10] and asphyxia and hypothermia might also cause fluctuation of cerebral blood flow [11], being possible risk factors for intraventricular hemorrhage (IVH) [12]. Performing magnetic resonance imaging (MRI) in combination with proton magnetic resonance spectroscopy (MRS) is a widely accepted imaging method for quantifying the extent of hypoxic ischemic brain injury, and predicting outcome [13‐24]. Involvement of the posterior limb of the internal capsule (PLIC) was found to be associated with abnormal motor outcome [13, 23]. Early death and the most severe motor impairment as cerebral palsy were seen in association with basal ganglia injuries [23]. In a meta-analysis including thirty-two studies and 860 infants with neonatal encephalopathy deep gray matter Lac/NAA was found to be the most accurate quantitative MR biomarker for prediction of neurodevelopmental outcome [22]. Yet, the different types of ICHs visible on MRI in asphyxiated neonates, were not considered in the previous studies, although they may be of additional impact on the long term outcome of neonatal HIE, as they may carry a potential risk of intracranial complication when hypothermic treatment is adopted [9]. The many previous studies emphasizing the prognostic effect of signal abnormalities on conventional sequences, diffusion restriction pattern, and brain MRS either analyzed a relatively small sample size or MRS was performed after the first week of life or was not involved in the evaluation, meaning that those HIE positive infants having only spectroscopic abnormality were not involved in the study population [13, 15, 25‐27].

In this retrospective observational study of cooled infants with the clinical diagnosis of HIE, we aimed to investigate whether the coexistence of ICH on early MRI with the imaging signs of HIE (abnormal MRS and/or typical patterns of diffusion restriction) have an impact on the prognosis.

Clinical characteristics of 108 neonates with moderate-to-severe HIE meeting the eligibility criteria and 102 patients excluded from the study are presented in Table 1. Clinical parameters of the included and excluded group of patients showed no significant difference.

The average age at which brain MRI scans were obtained was 65.12 ± 46.34 h. All brain MRIs were performed after passive cooling started during transport or after the initiation of therapeutic hypothermia. Of these 108 patients 9 patients died in the perinatal period (6.67%). None of our patients died between 28 days and the neurodevelopmental follow-up examination. Twelve children had significantly delayed development (MDI and/or PDI < 70). Seven of these patients were diagnosed with cerebral palsy. Twenty-seven children received an MDI and/or PDI score compatible with mildly delayed development (70–84).

The 108 subjects were divided into four groups based on the presence (+) or absence (-) of the MRI signs of ICH (Fig. 2) and HIE (Figs. 3 and 4): Group1: Infants presenting neither the signs of HIE nor ICH (Additional file 2). Group2: Patients without the imaging signs of HIE showing ICH (Additional file 3). Group3: Both HIE and ICH is visible (Additional file 4). Group4: Only HIE signs are present (Additional file 5). Neurodevelopmental outcome results were registered for each patient.

Here, we assessed the rate of ICH in a cohort of patients with clinical signs of HIE, and also considered the effect of ICH on the long term outcome in this patient population. Several studies demonstrate the benefits and compare the accuracy of MRI in identifying HIE and predicting the long term neurodevelopmental outcome [1, 36, 40‐42], nevertheless, to our knowledge this was the first study considering the possible relevance of ICH accompanying HIE. Three major randomized controlled trials as the NICHD study [3], the TOBY trial [4] and the ICE trial [6, 7] have demonstrated a reduction in death and neurological impairment after hypothermia treatment. The above trials also included brain MRI, but compared to our study these were relatively late examinations performed after the first week of life in average, assessing the patterns of injury classified by Rutherford mainly on conventional T1- and T2-weighted sequences, and partially on diffusion weighted imaging. The average age on completion of MRI in our study was around the third day of life. The pattern of the injury defined as predominantly basal ganglia or thalamic lesions, watershed predominant injury or global brain insult were classified on diffusion weighted images in most of the cases. MR-spectroscopy was also considered, revealing 29 patients with spectra compatible with HIE and without other imaging signs of hypoxic-ischemic injury.

Published incidence of asymptomatic ICH after vaginal delivery of full-term neonates is approximately 25%, [43] yet symptomatic ICH is much less common, accounting for 4 per 10,000 live births [44]. The rate of ICH in our patient population of cooled asphyxiated infants was higher (36%) than found in literature. The majority of ICHs in our study were small extra-axial or petechial parenchymal hemorrhages localized in the posterior fossa, but there also were multifocal hemorrhages with various patterns of bleeding. The origin of the observed hemorrhages may be related to delivery [45] or to the cooling itself [11, 12], and while the cooling-induced thrombocytopenia [6] may potentially lead to a progression of these bleedings, we have no evidence of progression in our data, partly due to a lack of follow-up imaging. Nevertheless, progression seems to be unlikely given the fact that these ICHs proved to have no significant modulatory effect on the long term mental (MDI) and psychomotor (PDI) developmental outcome.

Early MRI evidence of HIE proved to be a significant predictor of an adverse outcome eventuating more than six times higher odds for abnormal mental development in an asphyxiated child compared to a neonate without the imaging proof of HIE with or without the presence of ICH. Vice versa, in almost all of the cases clinically categorized as HIE (n = 30) but lacking MRI evidence of HIE the neurodevelopmental outcome was within normal limits. This is in line with the converging evidence showing the negative outcome associated with the presence of HIE signs on MRI imaging performed after the first week of life [3, 4, 6, 7], and also consistent with the literature regarding early MRI [34].

There are some aspects of the study that can be considered as limitations, the first is the retrospective nature of the data presented. A prospective trial could allow for better controlled patient selection, and could also lead to more consistent data acquisition, nevertheless clinical data obtained in a general setting is equally important [35]. While our data derived from 210 consecutive patients gives a cross sectional overview of the patient population we observed in a 9-year period, it has its own inconsistencies regarding imaging and follow-up, eventually leading to a relatively high rate of exclusion. This may be considered a limitation of our study, nevertheless the lack of significant differences in the basic clinical parameters (e.g. Apgar scores, pH, etc.) between the included and excluded patient cohorts suggests that our sampling is representative of the population categorized as HIE patients based on clinical grounds, i.e. the TOBY criteria [4].

It is important to note, that we excluded those patients whose first MRI was performed after the first week of life, given the main body of HIE-related literature [34, 38, 40] this can be considered as a shortcoming, however our goal was to investigate the early and transient signs of HIE on MRI alongside with the presence of ICH. The fact that our findings on the outcome-modifying effects of HIE signs present on MRI are in line with the literature obtained on later acquired imaging data [3, 4, 6, 7] justifies the appropriateness of our selection criteria.

The relatively wide age range at follow-up developmental testing can also be considered as a potential limitation, however the BSID-II is standardized to be used between 1 and 42 months of age, and the patient’s age is properly taken into consideration for the final scores [32, 46]. Furthermore, one may also consider the use of BSID-II instead of BSID-III a limitation, please note however that the BSID-III was not translated and standardized to Hungarian until 2018; till then the BSID-II was the standard developmental test battery available in Hungarian.

The lack of an objective scaling system to image the severity of HIE may also be considered as limitation. Although, there are several scoring systems available, most of the scales depend on late MRI and conventional images, some of them on the early diffusion pattern alone, yet there is no score including MR spectroscopy in the grading, which could be used in our evaluation [27, 37, 47‐49]. The reason for the lack of published MRS grading is that there is very narrow time window for the initiation of the hypothermia (maximum of 6 h from birth) [50, 51], which limits the opportunity of performing an early MRI and there are still many centers where MR-spectroscopy is not included in the imaging protocol [52, 53]. The recent recommendation of the American College of Obstetricians and Gynecologists outlines the importance of early MRI (first 24–96 h of life) for likely timing of the hypoxic-ischemic insult and further follow-up imaging (7–21 days of life) to define the full nature of the abnormalities [54], but does not define or suggest a protocol for brain MRI in an asphyxiated infant.

Finally, one can consider the fact that MR examinations in our study were performed after passive cooling during transport or under or after hypothermic treatment as a limitation of our study, however we do not consider this a clinically relevant limitation as there is published evidence that the predictive value of MRI for subsequent neurological impairment is not affected by previous hypothermia [55]. In fact, MRI is accurate for evaluating the presence and extent of hypoxic brain injury under and after hypothermia [56].

To successfully overcome the above-mentioned limitations more studies on early MRI need to be performed involving more subjects and gathering more imaging data to gain better insight whether there is any interference between hypothermic treatment and imaging that may render the MRI results of asphyxiated infants unremarkable.

This study attempted to determine the predictive value of intracranial hemorrhage in association with HIE related MR imaging signs in comparison with neurodevelopmental results in cooled asphyxiated infants. Intracranial hemorrhage took a remarkable share affecting approximately one-third of infants treated with hypothermia. Intracranial hemorrhagic complications which were present before or evolve during treatment showed no significant influence on outcome despite the known negative hematological (platelet dysfunction, thrombocytopenia) effects of cooling. Having more than six time higher odds for abnormal neurodevelopmental outcome the presence of HIE related MR-spectroscopy abnormalities and diffusion restriction proved to have important prognostic value. Therefore, the implementation of early MRI in the diagnostic algorithm needs to be considered.