Hypothermia-treated neonates with hypoxic-ischemic encephalopathy: Optimal timing of quantitative ADC measurement to predict disease severity

Abstract

To determine the optimal time window for MR imaging with quantitative ADC measurement in neonatal HIE after hypothermia treatment, a retrospective review was performed on consecutive hypothermia-treated term neonates with HIE, with an initial and follow-up MR imaging within the first two weeks of life. Three neuroradiologists categorized each set of MR imaging as normal, mild, moderate or severe HIE based on a consensus review of the serial imaging. The lowest ADC values from the white matter, corpus callosum, and basal ganglia/thalamus were measured. The ADC values between mild-moderate and severe HIE were compared using a Student’s t-test over a range of different time windows. A total of 33 MR imaging examinations were performed on 16 neonates that included three normal, four mild, five moderate, and four severe HIE. The time window of 3–10 days showed a statistically significant decrease in ADC value in severe HIE compared to mild-moderate HIE in all three locations, respectively: white matter 0.5 ± 0.22 versus 0.83 ± 0.27 (p value 0.01), corpus callosum 0.69 ± 0.19 versus 0.91 ± 0.17 (p value 0.01), and basal ganglia/thalamus 0.63 ± 0.16 versus 0.98 ± 0.06 (p value <0.01). The range of 3–10 days is the optimal time window for MR imaging with quantitative ADC after hypothermia treatment.

Introduction

Hypoxic ischemic encephalopathy (HIE) is estimated to affect 3.75 per 1000 full-term live births and carries a high risk of morbidity and mortality.^1–3 In the last decade, hypothermia has been shown to reduce mortality and is increasingly used as the treatment of choice.^4,5 Magnetic resonance (MR) imaging also has been increasingly used to provide confirmation of HIE and predict long-term outcome based on the distribution of injury: A watershed predominant brain injury is correlated with mild disease, while a basal ganglia/thalamus predominant brain injury has been correlated with a greater severity of disease.^6,7 Brain injury involving the watershed zone, cortex, and basal ganglia/thalamus carries the worst prognosis with the highest rate of unfavorable outcomes.^6–9

Diffusion-weighted imaging (DWI) with quantitative apparent diffusion coefficient (ADC) measurement is useful for predicting disease severity.^8–12 Restricted diffusion of specific locations, such as the corpus callosum, has been useful in predicting disease outcome.¹³ One of the major advantages of DWI is the increased sensitivity for HIE compared to T1-weighted imaging (T1WI) and T2-weighted imaging (T2WI) during the first few days of life.¹⁴ HIE brain injury may appear earlier on DWI prior to the full extent of brain injury development on T1WI and T2WI, and therefore, DWI has the potential for earlier in-hospital imaging assessment and may aid in clinical decision making regarding long-term prognosis. After reaching a nadir, the decreased ADC reverts back toward normal or “pseudonormalize” around six to eight days in neonates without hypothermia treatment.^8,15,16

Understanding the impact of hypothermia on MR imaging is crucial given the important role of both hypothermia and MR imaging in the care of neonates with HIE. Prior studies in neonates with HIE have shown that hypothermia reduces the frequency of MR findings, increases the measured ADC value, and delays the time at which ADC pseudonormalizes.^17–21 However, there is still limited understanding regarding the optimal time window for MR imaging with quantitative ADC after hypothermia treatment. Achieving this end is essential for coordinating the proper timing of MR imaging as neonates with HIE are often critically ill. Consequently, we quantified ADC values and compared to HIE severity over a range of different time windows to determine when quantitative ADC measurement is predictive of disease severity.

Materials and methods

Patient selection

A retrospective review was performed on a total of 16 consecutive neonates who underwent hypothermia treatment for clinically diagnosed HIE at Santa Clara Valley Medical Center from 2008 to 2012. Inclusion criteria were based on both physiologic and neurologic parameters as outlined previously in the literature.^22,23 Neonates meeting the criteria underwent whole-body cooling for 72 hours within the first six hours of life following published protocols.^22,23 At the ending of the cooling period, each neonate was subsequently re-warmed over 24 hours. Each neonate underwent MR imaging without sedation at 3–14 days. MR imaging performed before three days was not included in the study because HIE is known to be masked or manifest variably during this early time period.²⁴

Image acquisition

All neonates were examined with either a 1.5T or 3T General Electric MR scanner. Sequences included axial DWI, axial T2, axial T2 fluid-attenuated inversion recovery (FLAIR), axial gradient echo (GRE), axial T1 fast-spin echo (FSE), and sagittal T1 FLAIR or T1. Post-processed ADC maps were created from the DWI sequences.

Image interpretation

Each set of initial and follow-up MR imaging from each neonate was evaluated by three board-certified neuroradiologists and achieved a consensus interpretation as normal, mild, moderate or severe HIE based on the distribution of injury on all MR sequences (Figure 1). Normal was assigned to each set of initial and follow-up MR imaging if both examinations had no evidence of brain injury on all sequences. Mild HIE was assigned to each set of MR imaging if there was brain injury involving the white matter predominantly in the watershed zone on either the initial or follow-up examination. Moderate HIE was assigned to each set of MR imaging if there was brain injury predominantly in the basal ganglia and/or thalamus on either the initial or follow-up examination. Severe HIE was assigned to each set of MR imaging with brain injury involving the watershed zone, cortex, and basal ganglia/thalamus on either the initial or follow-up examination. The presence of ischemic sequelae including encephalomalacia, parenchymal hemorrhage, and intraventricular hemorrhage was also determined by consensus and recorded, although this was not used as part of the grading (Figure 2).

Figure 1.

(a) Examples of mild HIE (white matter predominant injury): restricted diffusion involving the corpus callosum and white matter (left, axial DWI), and restricted diffusion involving only the white matter (right, axial DWI). (b) Examples of moderate HIE (basal ganglia/thalamus predominant injury): signal abnormality involving the basal ganglia (left, axial FLAIR), and restricted diffusion involving the left frontal white matter, right thalamus and splenium of corpus callosum (right, axial DWI). (c) Examples of severe HIE (total brain injury): restricted diffusion involving the white matter, cortex, corpus callosum, and basal ganglia/thalamus (left and right, axial DWI). HIE: hypoxic-ischemic encephalopathy; DWI: diffusion-weighted imaging; FLAIR: fluid-attenuated inversion recovery.

Figure 2.

(a) Example of ischemic sequelae (axial T2WI): bilateral cerebral parenchymal hemorrhage and diffuse encephalomalacia. (b) Example of ischemic sequelae (axial T2WI): intraventricular hemorrhage within the occipital horn of right lateral ventricle. T2WI: T2-weighted imaging.

ADC measurement

While the consensus on HIE severity grading was based on distribution of abnormality as described above, separately, the ADC value was quantitatively measured. A trainee radiologist, blinded to the results of the consensus interpretation, measured ADC values on post-processed ADC maps by using a 3–4 mm diameter region of interest (ROI) at the following locations: white matter, corpus callosum, and basal ganglia/thalamus. The ROIs were reviewed by two of the above-mentioned board-certified neuroradiologists to confirm proper placement. For each initial MR imaging, the lowest achievable ADC value in each location was recorded. For all follow-up MR imaging, ADC value at the same location as the initial examination was measured.

Since ADC values vary regionally in the normal neonatal brain and across different MR scanners, the ADC values from HIE MR imaging were normalized to the mean ADC values from all normal MR imaging from the corresponding location as previously described in the literature.^17,25,26 First, the mean normal ADC value was calculated from all the normal sets of MR imaging for each of the three locations: white matter, corpus callosum, and basal ganglia/thalamus. Then each ADC value from HIE MR imaging was normalized by dividing by the mean normal ADC value from the respective locations.

Statistical analysis

Each set of MR imaging that was not in the normal group was divided into two groups based on the consensus interpretation: those with mild-moderate HIE (mild HIE or moderate HIE) and severe HIE. The mild and moderate HIE were combined for the analysis because of the small number of examinations. Student’s t-test was used to compare the normalized ADC values between mild-moderate and severe HIE in each of the three anatomical locations (white matter, corpus callosum, and basal ganglia/thalamus) over 3–7, 3–10, 7–14, and 10–14 days. Also, all normalized ADC values were plotted against time to determine whether the ADC pseudonormalization trend described in previous studies also applies to our population.¹⁷ Statistical analysis was completed using a software package (Stata version 8.0; StataCorp, 2003, College Station, TX). For all statistical tests, all p values less than 0.05 were designated to be significant.

Results

Demographics and imaging findings

The study included 16 neonates with three normal, four mild, five moderate, and four severe HIE (Table 1). Gestational age and gender were proportionally distributed across each group (Table 1). The majority of HIE MR imaging demonstrated restricted diffusion or signal abnormalities on the initial MR imaging (Table 2) with improvement of the brain injury on the follow-up MR imaging. Additional injury developed in a number of follow-up HIE MR imaging (MO1, MO2, MO5, S2, and S3) (Table 2, Figures 3–4) although this did not change the HIE severity category of the examination. Overall, a greater frequency of the severe HIE demonstrated ischemic sequelae compared to the mild-moderate HIE: Encephalomalacia was seen in three out of four severe HIE (75%) compared to none (0%) of the mild-moderate HIE (Table 2), and parenchymal hemorrhage was seen in one out of four severe HIE (25%) compared to one out of nine (11%) of the mild-moderate HIE (Table 2). Encephalomalacia developed associated with parenchymal hemorrhage in one of severe HIE (Figure 2(a)). Other ischemic sequela such as intraventricular hemorrhage was also noted (Figure 2(b)).

Table 1.

Summary of the demographic and clinical history of all neonates.

Findings on all MR sequences	Normal	Mild HIE	Moderate HIE	Severe HIE
Total (n)	3	4	5	4
Gestational age (weeks)	40	40.4	38.5	39.3
Mean range	37.3–42	39.9–41.1	36.6–40.1	36–42.1
Gender
Female	1	3	3	2
Male	2	1	2	2

MR: magnetic resonance; HIE: hypoxic-ischemic encephalopathy.

Table 2.

Summary of MR findings.

Severity	Neonate	Age (days) and findings	Ischemic sequelae
Mild	MI1	6 White-matter injury 13 Normal	No
	MI2	4 Corpus callosum injury 9 Persistent injury of the corpus callosum	No
	MI3	7 Subtle signal abnormalities of the white matter	No
	MI4	7 White-matter injury	No
Moderate	MO1	4 Abnormalities in the basal ganglia/thalamus and corpus callosum 10 Development of additional abnormalities in the white matter	Yes
	MO2	6 Abnormalities in the basal ganglia/thalamus, corpus callosum, and white matter 10 Development of additional abnormalities in the white matter	No
	MO3	4 Abnormalities in the basal ganglia/thalamus, corpus callosum, and white matter 9 Improvement of abnormalities	No
	MO4	4 Abnormalities in the basal ganglia/thalamus, corpus callosum, and white matter 13 Normal	No
	MO5	4 Subtle abnormalities in the basal ganglia/thalamus 9 Improvement of abnormalities in the basal ganglia/thalamus, but development of additional abnormalities in corpus callosum and white matter	No
Severe	S1	3 Extensive abnormalities in the basal ganglia/thalamus, corpus callosum, cortex, and white matter 7 Persistent abnormalities	Yes
	S2	1 Abnormalities in the basal ganglia/thalamus, corpus callosum, cortex, and white matter 6 Extensive development of additional abnormalities in the thalamus, cortex, and white matter 11 Persistent abnormalities	Yes
	S3	5 Extensive abnormalities in the basal ganglia/thalamus, corpus callosum, cortex, and white matter 10 Heterogeneous areas of improvement as well as development of additional abnormalities in the basal ganglia/thalamus, corpus callosum, and white matter	Yes
	S4	1 Extensive abnormalities in the basal ganglia/thalamus, corpus callosum, cortex, and white matter 4 Persistent abnormalities	Yes

MR: magnetic resonance.

Figure 3.

(a) Evolution of brain injury: initial MR imaging at four days (left, axial DWI) demonstrating restricted diffusion in the bilateral frontal lobes and corpus callosum, and subsequent MR imaging at nine days (right, axial DWI) demonstrating interval decrease in conspicuity or “pseudonormalization” of increased DWI signal abnormality in the bilateral frontal lobes and corpus callosum (white arrows). (b) Evolution of brain injury: initial MR imaging at one day (left, axial DWI) demonstrating minimal restricted diffusion in the splenium of the corpus callosum (black arrow), and subsequent MR imaging at four days (right, axial DWI) demonstrating interval development of extensive heterogeneous bilateral restricted diffusion. Although the initial MR imaging at one day (left, axial DWI) was not used for our statistical analysis, such an example underestimated the full extent of brain injury by MR imaging performed prior to three days. MR: magnetic resonance; DWI: diffusion-weighted imaging.

Figure 4.

(a) Spatial heterogeneity of brain injury: initial MR imaging at four days (left, axial DWI) demonstrating heterogeneous restricted diffusion involving the white matter, cortex, corpus callosum, and basal ganglia/thalamus. Subsequent MR imaging at nine days (right, axial DWI) demonstrating resolution of restricted diffusion in the right temporal lobe and left basal ganglia (white arrows) as well as progression of restricted diffusion in the left temporal lobe and right basal ganglia (black arrows). This heterogeneity did not change our categorization of disease severity. (b) Spatial heterogeneity of brain injury: initial MR imaging at four days (left, axial FLAIR image) demonstrating no significant brain injury. Subsequent MR imaging at 10 days (right, axial FLAIR image) demonstrating interval development of subtle punctate signal abnormalities in the right periventricular white matter (white arrows). MR: magnetic resonance; DWI: diffusion-weighted imaging; FLAIR: fluid-attenuated inversion recovery.

Time window

Severe HIE demonstrated a lower mean normalized ADC value, corresponding to a greater magnitude of restricted diffusion, compared to mild-moderate HIE (Table 3) with a statistically significant difference during 3–10 days in all three locations based on Student’s t-test. For three to seven days, however, a statistically significant decrease in normalized ADC value was demonstrated only in the basal ganglia/thalamus and not in the other two locations. When expanding the time frame from three to seven days to 3–10 days, there was not only an increase in sample size but also an increase in the magnitude of difference that both affect the T-statistic. There was less statistical significance between the two groups during the ranges of 7–14 or 10–14 days. The trend for decreased ADC value to revert back to normal over time, or pseudonormalization, was illustrated using scatter plots showing an initially decreased ADC value that then increased toward 1.0 (Figure 5(a)–(c)).

Table 3.

Comparison of normalized ADC values between mild-moderate and severe HIE during different time windows.

DOL	Location	Mild–Mod		Severe
		Mean ADC ratio	Std dev	Mean ADC ratio	Std dev	p value
3 to 7	White matter	0.71	0.30	0.53	0.24	0.13
	Corpus callosum	0.84	0.18	0.67	0.20	0.07
	Basal ganglia/thalamus	0.99	0.07	0.64	0.18	<0.01a
3 to 10	White matter	0.83	0.27	0.50	0.22	0.01a
	Corpus callosum	0.91	0.17	0.69	0.19	0.01a
	Basal ganglia/thalamus	0.98	0.06	0.63	0.16	<0.01a
7 to 14	White matter	0.89	0.18	0.79	0.36	0.27
	Corpus callosum	0.96	0.11	0.84	0.06	0.04a
	Basal ganglia/thalamus	0.97	0.05	0.84	0.27	0.08
10 to 14	White matter	0.89	0.18	0.72	0.48	0.26
	Corpus callosum	0.98	0.14	0.80	0.01	0.09
	Basal ganglia/thalamus	0.98	0.05	0.82	0.37	0.19

^aStatistical significance. DOL: day of life; ADC: apparent diffusion coefficient; HIE: hypoxic-ischemic encephalopathy; Mod: moderate.

Figure 5.

(a) Time course of normalized ADC value of the white matter in mild-moderate versus severe HIE. (b) Time course of normalized ADC value of the corpus callosum in mild-moderate versus severe HIE. (c) Time course of normalized ADC value of the ganglia/thalamus in mild-moderate versus severe HIE. ADC: apparent diffusion coefficient; HIE: hypoxic-ischemic encephalopathy.

Discussion

Understanding the optimal time window for MR imaging with quantitative ADC after hypothermia treatment is essential for coordinating the proper timing of MR imaging as neonates with HIE are often critically ill. It is already known that ADC value reaches a nadir during the first few days of life prior to pseudonormalization and reverts back to normal around six to eight days in neonates without hypothermia treatment.^8,15,16 In neonates with hypothermia treatment, a delay in pseudonormalization has been suggested although this has yet to be confirmed.¹⁷ For our study, we included only neonates with initial and follow-up MR imaging during the first two weeks of life in order to capture the ADC value changes when they were most apparent prior to pseudonormalization. Although we did not include data prior to three days because of clinical necessity for hypothermia treatment and re-warming prior to three days, the literature has suggested that earlier MR imaging may underestimate the full extent of brain injury.²⁴ We found that quantitative ADC during the time window of 3–10 days would be predictive of HIE disease severity after hypothermia treatment. During the time window of 3–10 days, severe HIE demonstrated a statistically significant decrease in ADC value than mild-moderate HIE. There was much less statistical significance during the ranges of 7–14 or 10–14 days, which matches our current understanding of pseudonormalization.¹⁷ There is likely an inflection point past seven days and somewhere between seven and 10 days although we were unable to determine a more precise inflection point because of the small number of patients.

The pseudonormalization of ADC value in our hypothermia-treated HIE neonates (Figure 5(a)–(c)) occurred between seven and 10 days, which is delayed compared to normothermic neonates in the literature in which normalization occurs between six and eight days.^8,15,16 Our finding of delayed ADC pseudonormalization in hypothermic neonates agrees with a similar study although an even longer time course at 11–12 days has been suggested.¹⁷ The delay in ADC pseudonormalization in hypothermia-treated HIE neonates compared to normothermic neonates in the literature suggests that hypothermia may have an effect on the evolution of brain injury. The mechanism by which hypothermia delays the evolution of HIE is unknown, although animal studies suggest that hypothermia may play a role in reducing secondary energy failure, release of excitatory amino acids such as glutamate, and apoptosis.^27,28

Several MR imaging (MO1, MO2, MO5, S2, and S3) in our study also suggested that HIE is an evolving and heterogeneous process in which temporal and regional differences likely occur with reperfusion brain injury (Figures 3–4). After the initial MR imaging, additional brain injury developed with areas of restricted diffusion and/or altered T1/T2 signal in the white matter, cortex, corpus callosum, and basal ganglia/thalamus. It is noteworthy that despite the evolving and heterogeneous manifestation of HIE on MR imaging, the interval changes between our initial and follow-up MR imaging did not change the HIE severity categorization. Prior studies have also noted this phenomenon, which may reflect the biphasic course of energy failure in HIE characterized by an initial predictable outcome of acute cell death followed by continued neuronal injury up to a matter of weeks and even months.^29–32 Because brain injury in asphyxiated neonates may be an ongoing process even after hypothermia, this raises the question of whether adjuvant therapies throughout the early post-natal period may provide additional benefits.³³

Since ADC values vary regionally in the normal neonatal brain and across different MR scanners, we normalized the HIE ADC values to the mean normal ADC values from the corresponding location as previously described in the literature.¹⁷ Also, we evaluated the frequency of ischemic sequelae, such as encephalomalacia and parenchymal hemorrhage, to corroborate the severity grading that was based on injury distribution shown on MR imaging because it was not feasible to perform clinical follow-up over a short period of time. Our study showed that severe HIE demonstrated a greater frequency of ischemic sequelae such as encephalomalacia and parenchymal hemorrhage. This finding is not surprising since prior studies that have evaluated clinical outcomes have demonstrated that a diffuse distribution of brain injury, as seen in our severe HIE, is associated with unfavorable outcomes such as death, cerebral palsy, and decreased cognitive scores at 2 years of age.^7,9,10

Small sample size is a limitation of our study, in part due to a desire to keep our study group homogenous with all neonates from the same institution with similar risk factors and receiving the same treatment during consecutive years. Given the small number of neonates and the overlap in disease between the mild and moderate HIE, we combined mild and moderate HIE for our statistical analysis. The distribution of brain injury in our neonates agrees with another larger study without hypothermia treatment.⁶ We were able to obtain statistically significant results in spite of a small sample size over several days of life ranges. However, the small sample size increases probability of a Type II error for our negative results at 7–14 and 10–14 days. Another potential limitation to the study was that there was no standardization of the timing of the MR imaging as this depended on whether the neonate was clinically stable enough for imaging. Finally, we did not evaluate long-term clinical outcomes to confirm HIE severity with the pattern of injury seen on the MR imaging in our study. However, our group plans to pursue a study addressing this question in the future when the neonates’ long-term clinical data will be available.

Clinical implications

Quantitative ADC is predictive of HIE severity after hypothermia treatment at 3–10 days. Early MR imaging immediately after hypothermia treatment and re-warming after three days may be helpful by providing early confirmation of the diagnosis, assessing HIE severity, guiding ongoing clinical care, or expediting hospital discharge. MR imaging up to 10 days may still provide critical information prior to ADC pseudonormalization given that neonates may be clinically unstable and preclude MR imaging during the first few days.

Conclusion

The optimal time window for performing MR imaging on neonates with HIE after hypothermia treatment is at 3–10 days. During this time window, a greater decrease in ADC value is associated with severe HIE. After this time window, the ADC ratio normalizes at 7–10 days, which is delayed compared to six to eight days reported with neonates without hypothermia treatment in the literature.^8,15,16

Conflict of interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.