Reader TTS

You have full access to this article via your institution. Download PDF We appreciate the comprehensive comments provided by Mehmet et al.1 regarding our study of brain MRI as a predictor of

outcome in infants with hypoxic-ischemic encephalopathy (HIE).2 Because MRI plays an important role in assessing the presence, location and severity of brain injury in HIE, at least 11

different brain injury classification systems have been published,3 each with its distinct advantages and disadvantages. A recent study of four highly rated classification systems,4,5,6,7

including the Trivedi system5 that was used in our study, reported AUC estimates ranging from 0.79 to 0.88 for predicting death or neurodevelopmental impairment (NDI).3 The NICHD

classification system similarly demonstrated an AUC of 0.77 for predicting death or severe disability in infants enrolled in the Late Hypothermia Trial.6 Upon further analysis of our HEAL

Trial data, the Trivedi system demonstrates an AUC of 0.73 (95% CI: 0.69–0.78) for predicting death or NDI. Although this estimate is slightly lower than the previous estimates summarized

above, such minor differences can be accounted for by differences in inclusion criteria and outcome definitions.3,6,8 Of note, this newly calculated AUC of 0.73 from the HEAL trial closely

aligns with the AUC of 0.72 that was originally reported in the Trivedi validation study.5 The Trivedi scoring system was designed to emphasize the central pattern of injury because this

pattern is known to be highly predictive of motor outcomes including cerebral palsy.9,10 However, as Mehmet et al. point out, the extra weight given to injury within the basal ganglia,

thalami, and posterior limb of the internal capsule could underestimate the prognostic importance of white matter injury. Furthermore, they point out that watershed distribution white matter

injury _alone_ without any evidence of central injury is unlikely to cause cerebral palsy unless the injury is extensive.11 Upon further analyses, we similarly found that among the 41

infants who had peripheral watershed injury but no central injury, five infants developed cerebral palsy, and all five had moderate to severe brain injury on MRI. As in previous studies,4

the two most common patterns of injury, i.e., central injury and peripheral watershed white matter injury, often co-exist in the HEAL cohort; of 111 infants who had peripheral watershed

injury, 70 (63%) also had evidence of central injury. It remains to be determined how best to combine information from these two overlapping injury patterns for purposes of prognostication.

Thoresen et al. found that the product of basal ganglia and white matter injury scores was more predictive of severe outcomes than either score alone.12 Weeke et al.4 found that white matter

sub-scores did not provide additional predictive value when added to a model that already included basal ganglia injury scores. In our study, we distinguished two different patterns of

white matter injury, and when we evaluated watershed distribution white matter injury separately from isolated punctate white matter lesions, only the former was associated with adverse

outcomes. A machine learning approach may help to determine which combination of individual brain MRI injury features will optimally improve our ability to predict outcomes. Furthermore,

although the Trivedi system measures the severity of injury in the posterior limb of the internal capsules,13 it does not measure injury to other parts of the corticospinal tract such as the

cerebral peduncles, as Mehmet et al. point out. In future work, we plan to perform advanced quantitative tract-based analyses to determine the effect of injury not just to these isolated

white matter structures, but to the entire corticospinal tract. The Trivedi and Weeke classification systems are similar, as both include injury severity measurements performed across

multiple regions of the brain. However, Mehmet et al. point out several limitations. Similar to other published HIE MRI scoring systems, the Trivedi system does not incorporate information

on intracranial hemorrhage4,6,7,12 or MRS data.6,7,12 The Trivedi system also does not measure injury to the hippocampi,14 and brainstem injury measurements are global as opposed to focused

on the cerebral peduncles. To address these limitations, we have established a collaborative group of experts to re-score the HEAL brain MRIs using an enhanced classification system that

will combine features of the Weeke and Trivedi systems while incorporating features that are novel to both systems such as hippocampal injury. We also plan to calculate interobserver

reliability when using this new scoring system, rather than relying on group consensus interpretation as was used in the current study. Mehmet et al. comment on another potential limitation

of the Trivedi system, namely that almost 50% diffusion restriction in the lentiform nuclei, bilateral posterior limb of the internal capsules, and bilateral cerebral peduncles could summate

to a score that falls within the “mild” category assuming that there is no concomitant injury seen on conventional T1 and T2 sequences. We agree that this would be a potential limitation if

we had acquired MRI data on the first day of age. However, our brain MRI studies were obtained at 4–6 days of age when signal abnormalities should have become visible on conventional

sequences. Thus, although the situation described by Mehmet et al. is theoretically possible, it is unlikely to be a significant problem during the time window when our neuroimaging studies

were performed. We thank Mehmet et al. for their insightful comments regarding the potential impact of image quality on our findings. The HEAL Neuroimaging Core used a harmonized MRI

protocol,15 provided training of site personnel to optimize scan quality, and actively monitored scan quality to address problems as they arose. As a result, there were only four infants

whose scans were excluded from the MRI study due to severe motion artifact, as already described in our publication. An additional 15 infants had mild, and three infants had moderate motion

artifact on at least one of their T1, T2 or DWI sequences. When we excluded these 18 infants, the predictive values reported in of our publication remained essentially unchanged (data

available upon request). We agree with Mehmet et al. that it is difficult to determine the timing of injury with any degree of certainty. We also agree with their statement that “observing

signal changes on conventional sequences (T1 and T2) but not on diffusion-weighted imaging suggests that the insult occurred some time before birth…” However, unlike the experience described

by Mehmet et al., this finding was not rare in the HEAL cohort. This discrepancy is most likely due to our subjects’ MRIs having undergone intense scrutiny by three neuroimaging experts for

research purposes. Indeed, many of the infants in our study had only minor signal changes on T1 or T2 sequences with no corresponding diffusion abnormalities, and these subtle findings were

not always noted in their clinical MRI reports. Whether such subtle abnormalities indicate subacute injury, transient vasogenic edema, or another process such as inflammation, and what

clinical significance they might carry, all remain open questions. It is important to emphasize that a brain MRI remains an important diagnostic and prognostic tool in the treatment of HIE,

even if the MRI injury score does not perfectly predict 2-year outcomes. As alluded to in the letter by Mehmet et al., a brain MRI is needed to rule out HIE mimics16 such as brain

hemorrhages, malformations, and other disorders that can cause neonatal encephalopathy. Our findings also confirm that the presence of severe injury, especially in the deep gray nuclei, is

highly predictive of poor outcome.7,9,10 For instance, the TOBY trial also found that a “major MRI abnormality” (i.e., moderate or severe basal ganglia/thalamic lesions, severe white matter

lesions, or an abnormal posterior limb of the internal capsule) was highly predictive of death or severe disability.6 The predictive value of neuroimaging data is likely to increase when

combined with clinical information, as was shown nicely by Thoresen et al.12 We agree with others7 that HIE prognostication should rarely be based on information obtained from a brain MRI

alone. In an on-going study, we are leveraging the power of machine learning to develop an optimal predictive model that combines brain MRI with neonatal EEG and other clinical markers of

disease severity. Because predictive models often perform differently in different populations,4 it is important to conduct independent evaluations of the predictive accuracy of such derived

models to avoid the problem of overfitting. In summary, although numerous studies describe statistically significant associations between brain MRI and HIE outcomes, measurement of MRI

brain injury alone is far from perfect in predicting the full range of long-term neurodevelopmental outcomes, regardless of the scoring system that is used. The AUC of the Trivedi score for

predicting poor outcome approximates that reported in other recent cohorts using alternative scoring systems. In clinical practice then, treating physicians must often provide families with

a range of possible outcomes. This approach not only reflects the current state of the literature, but also takes into account the plasticity of the neonatal brain and the importance of

environmental factors that will impact the developmental trajectory well beyond the newborn period.17 CHANGE HISTORY * _ 09 DECEMBER 2023 A Correction to this paper has been published:

https://doi.org/10.1038/s41390-023-02967-7 _ REFERENCES * Cizmeci, M. N., Martinez-Biarge, M. & Cowan, F. M. The predictive role of brain magnetic resonance imaging in neonates with

hypoxic-ischemic encephalopathy. _Pediatr Res_ https://doi.org/10.1038/s41390-023-02732-w (2023). * Wu, Y. W. et al. How well does neonatal neuroimaging correlate with neurodevelopmental

outcomes in infants with hypoxic-ischemic encephalopathy?. _Pediatr Res_ 94, 1018–1025 https://doi.org/10.1038/s41390-023-02510-8 (2023). * Langeslag, J. F. et al. Outcome Prediction and

Inter-Rater Comparison of Four Brain Magnetic Resonance Imaging Scoring Systems of Infants with Perinatal Asphyxia and Therapeutic Hypothermia. _Neonatology_ 119, 311–319 (2022). Article 

CAS  PubMed  Google Scholar  * Weeke, L. C. et al. A Novel Magnetic Resonance Imaging Score Predicts Neurodevelopmental Outcome After Perinatal Asphyxia and Therapeutic Hypothermia. _J.

Pediatr._ 192, 33–40.e2 (2018). Article  PubMed  PubMed Central  Google Scholar  * Trivedi, S. B. et al. A validated clinical MRI injury scoring system in neonatal hypoxic-ischemic

encephalopathy. _Pediatr. Radio._ 47, 1491–1499 (2017). Article  Google Scholar  * Rutherford, M. et al. Assessment of brain tissue injury after moderate hypothermia in neonates with

hypoxic-ischaemic encephalopathy: a nested substudy of a randomised controlled trial. _Lancet Neurol._ 9, 39–45 (2010). Article  PubMed  PubMed Central  Google Scholar  * Laptook, A. R. et

al. Limitations of Conventional Magnetic Resonance Imaging as a Predictor of Death or Disability Following Neonatal Hypoxic-Ischemic Encephalopathy in the Late Hypothermia Trial. _J.

Pediatr._ 230, 106–111.e6 (2021). Article  PubMed  Google Scholar  * Wu, Y. W. et al. Trial of Erythropoietin for Hypoxic-Ischemic Encephalopathy in Newborns. _N. Engl. J. Med_ 387, 148–159

(2022). Article  CAS  PubMed  PubMed Central  Google Scholar  * Martinez-Biarge, M., Diez-Sebastian, J., Rutherford, M. A. & Cowan, F. M. Outcomes after central grey matter injury in

term perinatal hypoxic-ischaemic encephalopathy. _Early Hum. Dev._ 86, 675–682 (2010). Article  PubMed  Google Scholar  * Martinez-Biarge, M. et al. Predicting motor outcome and death in

term hypoxic-ischemic encephalopathy. _Neurology_ 76, 2055–2061 (2011). Article  CAS  PubMed  PubMed Central  Google Scholar  * Martinez-Biarge, M. et al. White matter and cortical injury in

hypoxic-ischemic encephalopathy: antecedent factors and 2-year outcome. _J. Pediatr._ 161, 799–807 (2012). Article  PubMed  Google Scholar  * Thoresen, M. et al. MRI combined with early

clinical variables are excellent outcome predictors for newborn infants undergoing therapeutic hypothermia after perinatal asphyxia. _EClinicalMedicine_ 36, 100885 (2021). Article  PubMed 

PubMed Central  Google Scholar  * Rutherford, M. A. et al. Abnormal magnetic resonance signal in the internal capsule predicts poor neurodevelopmental outcome in infants with

hypoxic-ischemic encephalopathy. _Pediatrics_ 102, 323–328 (1998). Article  CAS  PubMed  Google Scholar  * Annink, K. V. et al. The long-term effect of perinatal asphyxia on hippocampal

volumes. _Pediatr. Res._ 85, 43–49 (2019). Article  PubMed  Google Scholar  * Wisnowski, J. L. et al. Integrating neuroimaging biomarkers into the multicentre, high-dose erythropoietin for

asphyxia and encephalopathy (HEAL) trial: rationale, protocol and harmonisation. _BMJ Open_ 11, e043852 (2021). Article  PubMed  PubMed Central  Google Scholar  * Parmentier CEJ, de Vries

LS, Groenendaal F. Magnetic Resonance Imaging in (Near-)Term Infants with Hypoxic-Ischemic Encephalopathy. _Diagnostics (Basel)_. 2022;12 https://doi.org/10.3390/diagnostics12030645 *

Benavente-Fernandez, I., Siddiqi, A. & Miller, S. P. Socioeconomic status and brain injury in children born preterm: modifying neurodevelopmental outcome. _Pediatr. Res._ 87, 391–398

(2020). Article  PubMed  Google Scholar  Download references FUNDING This study was funded by NINDS U01NS092764 and U01NS092553. AUTHOR INFORMATION AUTHORS AND AFFILIATIONS * Department of

Neurology, University of California San Francisco, San Francisco, CA, USA Yvonne W. Wu & Hannah C. Glass * Department of Pediatrics, University of California San Francisco, San

Francisco, CA, USA Yvonne W. Wu & Hannah C. Glass * Department of Radiology, Children’s Hospital Los Angeles, Los Angeles, CA, USA Jessica L. Wisnowski * Department of Pediatrics,

Children’s Hospital Los Angeles, Los Angeles, CA, USA Jessica L. Wisnowski * Department of Epidemiology, University of California San Francisco, San Francisco, CA, USA Hannah C. Glass *

Department of Pediatrics, Saint Louis University School of Medicine, St. Louis, MO, USA Amit M. Mathur * Department of Radiology, University of California San Francisco, San Francisco, CA,

USA Yi Li * Department of Biostatistics, University of Washington, Seattle, WA, USA Sarah E. Monsell & Robert C. McKinstry * Department of Pediatrics, University of Washington School of

Medicine, Seattle, WA, USA Sandra E. Juul * Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, MO, USA Robert C. McKinstry Authors * Yvonne W. Wu View

author publications You can also search for this author inPubMed Google Scholar * Jessica L. Wisnowski View author publications You can also search for this author inPubMed Google Scholar *

Hannah C. Glass View author publications You can also search for this author inPubMed Google Scholar * Amit M. Mathur View author publications You can also search for this author inPubMed 

Google Scholar * Yi Li View author publications You can also search for this author inPubMed Google Scholar * Sarah E. Monsell View author publications You can also search for this author

inPubMed Google Scholar * Sandra E. Juul View author publications You can also search for this author inPubMed Google Scholar * Robert C. McKinstry View author publications You can also

search for this author inPubMed Google Scholar CONTRIBUTIONS Y.W.W. made substantial contributions to conception and design, acquisition of data, and analysis and interpretation of data;

wrote the first draft of the article; and provided final approval of the version to be published. S.E.M. made substantial contributions to analysis and interpretation of data; revised the

article critically for important intellectual content; and provided final approval of the version to be published. H.C.G. made substantial contributions to conception and design and

interpretation of data; revised the article critically for important intellectual content; and provided final approval of the version to be published. J.L.W. made substantial contributions

to conception and design, acquisition of data; revised the article critically for important intellectual content; and provided final approval of the version to be published. A.M.M. made

substantial contributions to conception and design, acquisition of data; revised the article critically for important intellectual content; and provided final approval of the version to be

published. Y.L. made substantial contributions to acquisition of data; revised the article critically for important intellectual content; and provided final approval of the version to be

published. R.C.M. made substantial contributions to conception and design, acquisition of data; revised the article critically for important intellectual content; and provided final approval

of the version to be published. S.E.J. made substantial contributions to conception and design, acquisition of data, and analysis and interpretation of data; revised the article critically

for important intellectual content; and provided final approval of the version to be published. CORRESPONDING AUTHOR Correspondence to Yvonne W. Wu. ETHICS DECLARATIONS COMPETING INTERESTS

The authors declare no competing interests. ADDITIONAL INFORMATION PUBLISHER’S NOTE Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional

affiliations. The original online version of this article was revised. The following two references were missing in the original article: 1. Cizmeci, M.N., Martinez-Biarge, M. & Cowan,

F.M. The predictive role of brain magnetic resonance imaging in neonates with hypoxic-ischemic encephalopathy. Pediatr Res (2023). https://doi.org/10.1038/s41390-023-02732-w. Wu, Y.W.,

Monsell, S.E., Glass, H.C. et al. How well does neonatal neuroimaging correlate with neurodevelopmental outcomes in infants with hypoxic-ischemic encephalopathy?. Pediatr Res 94, 1018–1025

(2023). https://doi.org/10.1038/s41390-023-02510-8 They have both been cited in the first sentence of the article as follows: “We appreciate the comprehensive comments provided by Mehmet et

al. [1] regarding our study of brain MRI as a predictor of outcome in infants with hypoxic-ischemic encephalopathy (HIE).[2]” The other references were re-numbered accordingly. RIGHTS AND

PERMISSIONS Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE Wu, Y.W., Wisnowski, J.L., Glass, H.C. _et al._ Advancing brain MRI as a prognostic indicator in hypoxic-ischemic

encephalopathy. _Pediatr Res_ 95, 587–589 (2024). https://doi.org/10.1038/s41390-023-02786-w Download citation * Received: 19 July 2023 * Accepted: 02 August 2023 * Published: 11 September

2023 * Issue Date: February 2024 * DOI: https://doi.org/10.1038/s41390-023-02786-w SHARE THIS ARTICLE Anyone you share the following link with will be able to read this content: Get

shareable link Sorry, a shareable link is not currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt content-sharing initiative