ava i lab le a t www.sc iencedi rec t .com

journa l homepage : www.e l sev i er . com/ loca te /s iny

Seminars in Fetal & Neonatal Medicine (2008) 13, 30e34

Current controversies in hypothermicneuroprotection

John D.E. Barks*

C. S. Mott Children’s Hospital, University of Michigan Health System, Ann Arbor, MI 48109, USA

KEYWORDSHypoxiceischaemicencephalopathy;Neonatalencephalopathy;Newborn infant;Perinatal asphyxia;Therapeutichypothermia

Summary In 2005, three randomised controlled trials (RCTs) showed that treating infantswith hypoxiceischaemic encephalopathy (HIE) with hypothermia decreased the combined out-come of death or disability at 12e18 months, although treatment effects were modest. Morerecently, the US Food and Drug Administration (FDA) approved a device for selective head cool-ing. In addition, the protocol from another of the three trials, using equipment available inmany hospitals, has been in the public domain for over a year. Why has this not led to a con-sensus that hypothermia is the standard of care for HIE? This is explored. Important questionsfor future research will focus on ways to improve on initial results with cooling, such as drugplus hypothermia combination therapy and refining duration and depth of cooling or durationof rewarming. Although the latter are important questions for future clinical trials, those whoare convinced by the evidence to date should focus on safe implementation of cooling usingprotocols with established safety and efficacy and should consider ways to increase accessto cooling for eligible babies.ª 2007 Published by Elsevier Ltd.

Introduction

More than 2 years have passed since the results of threerandomised controlled trials (RCTs), which individually andcollectively demonstrated the efficacy of hypothermia forthe treatment of perinatal hypoxiceischaemic encephalop-athy (HIE) in term or late preterm infants, were publish-ed.1e4 Yet, to date, there is still no general consensus thathypothermia is the standard of care for such infants. Why isthis the case? This review aims to address some of the cur-rent controversies surrounding hypothermia as a treatmentfor HIE.

* Tel.: þ1 734 763 4109; fax: þ1 734 763 7728.E-mail address: jbarks@med.umich.edu

1744-165X/$ - see front matter ª 2007 Published by Elsevier Ltd.doi:10.1016/j.siny.2007.09.004

Why is everyone not convinced?

Until recently, the most definite sign that the neonatalcommunity was not universally convinced of the efficacyof hypothermia was that an RCT (the ICE trial), with anon-hypothermic control group, continued to enroll pa-tients in several countries. It would be logical to supposethat investigators who participated in the published orcompleted trials are most likely to be convinced of thesafety and efficacy of cooling, and that this convictionwould be demonstrated by their centres now offeringhypothermia to infants meeting their original trial criteria.This appears to be true for the National Institute of ChildHealth and Human Development (NICHD) and NeonatalResearch Network trial centres. Furthermore, many of thecentres in the Brain Cooling RCT also participated in the

Current controversies in hypothermic neuroprotection 31

subsequent FDA-approved continued access protocol andare now transitioning to the commercial Cool Cap� device.

One commonly expressed reservation about the earlycompleted trials is that none of them was powered toaddress the death versus disability question, i.e. cliniciansmay still need to be convinced that cooling is not simplypermitting more babies to survive with severe handicaps,without increasing the number of non-handicapped survi-vors. It is likely that the results of the three pending RCTs(see below) will provide enough patients to evaluate sepa-rately the effects of hypothermia on death and on disabilityin survivors by using meta-analysis. A related concern is thatoutcome in later childhood, e.g. cognitive outcome at schoolage, is unknown; interestingly, other new therapies inneonatology, like surfactant or inhaled nitric oxide, havenot been held to such a high standard before becomingaccepted therapies. Both the NICHD trial and the BrainCooling trial are planning school-age follow-up evaluations.

For others, an important question is, ‘have enoughpatients been studied to be reasonably certain that hypo-thermia is really effective?’ In simple terms, the concern isthat combining early ‘smaller’ trials of a new therapy ina meta-analysis is similar to taking multiple ‘peeks’ at dataand prematurely stopping a prospective trial e it increasesthe chance of a type I error. One way of addressing this hasrecently been proposed, using the ‘optimal information size’concept to calculate the number of infants needed in pooledtrials to confirm that a treatment effect exists and thatestimates of the treatment effect size are reasonable.5 Inves-tigators in the Brain Cooling and NICHD trials would probablydispute their characterisation as ‘small’ given the time andeffort involved in accruing over 200 patients each, but infact compared to some adult clinical trials in conditions ofgreater prevalence, they are relatively small trials. The re-sults of three additional large RCTs that have closed enroll-ment (the Chinese selective head cooling trial, the TOBYtrial and the ICE trial) should provide sufficient numbers of pa-tients to address the concern about optimal information size.5

What are the practical barriers toimplementation?

At the time of writing, there is no consensus that hypo-thermia should be the standard of care for infants withsuspected HIE. The advice from authoritative sources thathypothermia should only be undertaken using establishedprotocols with known safety and efficacy is still sound.6,7

For those convinced by the published evidence, cost and/or limited equipment availability may be barriers to imple-mentation of cooling within an individual centre, while ge-ography may pose an additional barrier to implementationof neonatal cooling at a regional level. However, lack ofavailability of detailed cooling protocols should no longerbe an issue.

The Olympic Cool Cap� System for selective head coolingis the only FDA-approved device for treatment of perinatalHIE with hypothermia. It is relatively expensive comparedwith other cooling systems. Its availability is limited at thepresent time due to the manufacturer’s phased roll-out,which has been made necessary by the fact that thecommercial device, a substantial update of the clinical

trial device, has never been ‘tested’ on a human infantprior to its commercial release. The Cool Cap� Systemnow features a graphical, touch-screen, user interfacethat walks the user through the entire cooling process. Al-though the Cool Cap� system is quite expensive comparedto the Blanketrol cooling mattress, economic modelling in-dicates that, from a societal perspective, the Cool Cap�

system would be cost-effective under several different re-gional implementation strategies in the state of Massachu-setts.8 Despite its cost-effectiveness at a regional/societallevel, individual hospitals might be deterred from coolingby price concerns.

The NICHD body cooling protocol is publicly available onthe Internet as a written protocol at https://neonatal.rti.org/studies_hypothermia.cfm. Unlike the FDA-approvedhead cooling system, the equipment used in the NICHD pro-tocol, the Blanketrol II� Hypo/hyperthermia system (Cin-cinnati Sub-Zero), was not designed or FDA-approved forneonates. Thus, its use for treatment of HIE in neonatesconstitutes an ‘off-label’ use, and for some hospitals in par-ticularly litigious settings, this may be a deterrent to its usein the treatment of HIE. Yet, it might be argued that manymedications commonly used in neonatology are used ‘off-label’ without hesitation. Although many neonatologistsmay not be familiar with the Blanketrol�, it is likely to befound at least in limited numbers in the operating roomsand/or patient equipment departments of many NorthAmerican hospitals and it is sold in other parts of the world,including Europe and the Middle East. Thus, some neonatalintensive care units (NICUs) would be able to implement theNICHD cooling protocol after investing the time necessaryto find and connect the equipment and supplies within theirown institution, and adapting the NICHD protocol for theirown use (e.g. creating a graphic presentation of the proto-col), as we have done at the University of Michigan HealthSystem. Others would need to purchase at least two of thedevices, one for the patient and one for back-up. It must beemphasised that safe use of the Blanketrol� device for hy-pothermia in neonates depends on strict adherence to theNICHD protocol, including use of a second, adult-sized mat-tress suspended from a pole to damp out temperature fluc-tuations that would otherwise occur with use of a systemoverpowered for neonates. The equipment setup and theprotocol are sufficiently complicated that body coolingshould not be undertaken on the spur of the moment, butonly after advance planning.

In Europe, the TOBY trial hypothermia protocol is in thepublic domain on the Internet, but the hypothermia deviceused was not available for purchase at the time of writing,and safety and efficacy data have not yet been reported.

Geographical barriers to cooling are relevant in someregions, because there is a consensus amongst all publishedprotocols that cooling needs to be initiated within 6 h ofbirth. This 6-h limit in clinical trials was based on animaldata, which suggest that the effectiveness of cooling dimin-ishes as time increases from the hypoxiceischaemic insult toinitiation of cooling, with the closing of the ‘window’ ata time between 5.5 and 8 h after the insult.9 In many largercountries, babies may be born at great distances fromregional NICUs, so that they could not possibly reach a‘cooling centre’ within 6 h of birth. If such infants are to ben-efit from therapeutic hypothermia, practical strategies will

32 J.D.E. Barks

be needed to evaluate and initiate cooling in the birth hos-pital, in collaboration with a regional referral NICU and priorto neonatal transport. Such strategies were implemented inthe published trial of Eicher et al., which demonstratedoverall efficacy1,10; cooling, prior to and/or during trans-port, was also incorporated into the TOBY and ICE trial pro-tocols. Until the results of the latter two trials arepublished, individual centres are left to consider whetherthe results of the Eicher trial and the experimental evidenceare compelling enough to support implementation of a trans-port cooling protocol and, as we have done, they maychoose to adapt Eicher’s transport cooling protocol to fittheir centre’s cooling strategy. Given the limited evidence,safety concerns should be paramount in developing proto-cols for initiation of cooling at birth hospitals without NICUsand/or during neonatal transport. In particular, there mustbe a method of safely measuring core (oesophageal or rec-tal) temperature; standard bedside digital thermometersmay not register temperatures in the target range for theestablished cooling protocols.

An often-overlooked barrier to cooling is denial of, orfailure to recognise, an infant’s birth depression and/orencephalopathy. Some practitioners delivering babies per-sist in paradoxically responding to a litigious environmentby not drawing umbilical cord blood gases, in a misguidedeffort to ‘protect’ themselves. This omission eliminatesvaluable information that could identify an infant requiringfurther evaluation for cooling, or that might suggest a non-hypoxiceischaemic explanation for neonatal depression.Although not every infant �36 weeks’ gestation with a cordpH< 7 is a candidate for cooling, all need evaluation forsigns of encephalopathy. Sometimes, evaluation for enceph-alopathy and consideration for cooling may be overlooked inthe flurry of managing other critical complications associ-ated with birth depression, such as hypotension, bloodloss, meconium aspiration syndrome and persistent pulmo-nary hypertension.

Even if the need to evaluate for encephalopathy isrecognised, clinicians vary in their level of experienceand confidence in performing the neonatal neurologicalexamination, and the signs of encephalopathy can besubtle, may not always fit neatly into published gradingschemes, can evolve over time, or may be mistaken for theside-effects of sedatives or of severe systemic illness.11 Am-plitude-integrated EEG (aEEG), which was originally used inthe Cool Cap trial to exclude infants anticipated to do wellwithout intervention, might become a useful way of decid-ing whether an infant with equivocal findings on examina-tion is ‘encephalopathic enough’ to benefit from cooling.

Who should be cooled? Who should not becooled?

The results of the Cool Cap trial indicated that the subgroupof infants with the most severe aEEG tracings (severesuppression plus seizures) did not improve with cooling,whereas the majority of infants, who had less severe aEEGabnormalities, showed improved outcome with cooling.2 Incontrast, a post hoc analysis from the Cool Cap trial,12 consis-tent with the results of the NICHD body cooling trial,3 showedthat clinical encephalopathy grade (Sarnat II versus III) did

not influence the effect of cooling on outcome. In a posthoc analysis, data from the NICHD body cooling trial were ex-amined to determine whether there were factors identifi-able prior to 6 h of life that could be incorporated intoeither a scoring scheme or a decision tree scheme to deter-mine which infants were likely to benefit from cooling andwhich would not.13 Although both approaches yielded meansof identifying infants in the NICHD study population thatwould not benefit from cooling, very few infants (2e3%)had such scores, and neither scheme has yet been validatedprospectively in a new patient population. Thus, these scor-ing systems are not currently suitable as a tool for excludinginfants from hypothermic intervention. Similarly, the BrainCooling investigators have not advocated use of aEEG criteriato exclude infants from cooling. To date, the most that canbe said, based on common sense and consistent with clinicaltrial protocols, is that infants judged to be in extremis arenot cooling candidates.

All currently published cooling protocols limit initiationof treatment to within 6 h of birth. There is no human datato support efficacy of cooling initiated >6 h after birth. Thismay be an important question for future study. However,the question could become largely irrelevant if recognitionof eligibility improves as evidence for, and acceptance of,cooling increase, and if protocols for evaluation and initia-tion of cooling at birth hospitals become more widelydisseminated.

How might we improve upon our results withcooling?

For those convinced by the current evidence, this is animportant question, which will be the focus for future re-search. Several possible strategies include combinations ofpotentially neuroprotective drugs with hypothermia; earlierinitiation of cooling; increased depth or duration of cooling;and prolonged rewarming. There is some experimentalevidence supporting additive or synergistic effects of drugswith hypothermia, including some FDA-approved drugs (top-iramate, N-acetylcysteine).14,15 Many other FDA-approveddrugs, e.g. anticonvulsants and anti-inflammatories, stillneed to be evaluated experimentally in combination withcooling, to provide a foundation for future human studies.

Although experimental evidence indicates that cooling ismore effective when initiated earlier after the insult,9 thetherapeutic time window for the benefit of cooling appearsto decrease as the severity of the initial insult increases.16

The two largest trials to date were unable to detect an ef-fect of time of initiation on outcome, perhaps becausecooling of most infants was started relatively late in thetime window in both trials, between 4 and 5 h after birth.2,3

In the Eicher trial, cooling was initiated with ice packs priorto or during transport, and enrollment time was thus earlierthan in the two larger trials1; it is tempting to speculatethat earlier time of initiation facilitated detection of effi-cacy, despite a smaller sample size than either the BrainCooling or the NICHD trial. Perhaps the results of the re-maining RCTs or analysis of data from the VermonteOxfordNeonatal Encephalopathy registry will shed additional lighton the importance of initiation time. There is no evidenceto indicate that longer durations of cooling or rewarming,

Current controversies in hypothermic neuroprotection 33

or greater depth of cooling will result in better outcomes(nor that they are safe), so the use of these approachesshould be limited to evaluation in RCTs.

Neurodevelopmental outcome

As indicated earlier, it is not yet known whether the earlypromising results with hypothermia based on evaluation at12e24 months will translate into improved performance atschool age. However, given the lack of any other therapiesfor HIE on the immediate horizon, this limitation is not likelyto present a barrier to implementation for many neonatol-ogists, nor is it likely to deter parents from accepting thetherapy.

Of more immediate concern, neonatologists and pediat-ric neurologists should remember that what we think weknow about the prognostic implications of early neurolog-ical findings and early neuroimaging, such as magneticresonance imaging (MRI) is based almost entirely on datafrom the pre-cooling era. As a neonatologist, who hasparticipated in the cooling of over 50 infants, using bothselective head cooling and the NICHD body cooling pro-tocol, the author has noted that even infants that ulti-mately do well at 18e24 months do not always showa dramatic improvement in neurological status after re-warming, but rather many seem to show a gradual recoveryover several days.

Regarding the prognostic implications of early MRI, e.g.at 7e10 days after birth, there is tantalising (but inconsis-tent) preliminary data to suggest that cooling is associatedwith a change in the pattern and/or extent of injury,17,18

but this has so far not been evaluated in large numbers ofpatients nor are there any reports so far of the correlationof early MRI findings in cooled infants with 18e24 months’neurodevelopmental outcome. It is possible that MRI find-ings at 7e10 days in infants with encephalopathy treatedwith cooling may represent reversible changes in evolution.The latter question is likely to be addressed by the NICHDtrial investigators in a secondary analysis. In the meantime,neonatologists and neurologists should exercise some cau-tion in assuming that what they knew about prognosticationin the pre-cooling era will still be true in infants that haverecently undergone therapeutic hypothermia, particularlyregarding infants with moderate encephalopathy or inter-mediate degrees of injury on MRI.

Conclusion

Until the results of several, now-closed, RCTs are pub-lished, the lack of consensus for therapeutic hypothermiaas the standard of care for infants with apparent HIE islikely to continue. There is no evidence that one method ofcooling is superior. Although devices and protocols toperform hypothermia according to already-published pro-tocols are now available, it is unclear how rapidly thistechnology will be disseminated while the results of theremaining trials are pending. It is reasonable for somecentres, particularly those that participated in the pub-lished trials, to offer cooling according to the publishedprotocols. It is also reasonable for centres that are not yetconvinced to await further evidence. In this interim period,

safety should be a priority, and data on cooled infantsshould be collected systematically, e.g. via the VermonteOxford Network Neonatal Encephalopathy registry or theTOBY Register.

References

1. Eicher DJ, Wagner CL, Katikaneni LP, et al. Moderate hypother-mia in neonatal encephalopathy: efficacy outcomes. PediatrNeurol 2005;32(1):11e7.

2. Gluckman PD, Wyatt JS, Azzopardi D, et al. Selective head cool-ing with mild systemic hypothermia after neonatal encephalop-athy: multicentre randomised trial. Lancet 2005;365(9460):663e70.

3. Shankaran S, Laptook AR, Ehrenkranz RA, et al. Whole-body hy-pothermia for neonates with hypoxiceischemic encephalopa-thy. N Engl J Med 2005;353(15):1574e84.

4. Edwards AD, Azzopardi DV. Therapeutic hypothermia followingperinatal asphyxia. Arch Dis Child Fetal Neonatal Ed 2006;91(2):F127e31.

5. Kirpalani H, Barks J, Thorlund K, Guyatt G. Cooling for neona-tal hypoxiceischemic encephalopathy e is the answer in? Pedi-atrics 2007;120(5):1131e2.

6. Higgins RD, Raju TN, Perlman J, et al. Hypothermia and perina-tal asphyxia: executive summary of the National Institute ofChild Health and Human Development workshop. J Pediatr2006;148(2):170e5.

7. Blackmon LR, Stark AR. Hypothermia: a neuroprotective ther-apy for neonatal hypoxiceischemic encephalopathy. Pediatrics2006;117(3):942e8.

Practice points

� Therapeutic hypothermia safely decreases thecombined outcome of death or disability at 12e24 months, in infants with apparent HIE.� Published protocols and the necessary equipment

are available.� Use of therapeutic hypothermia for HIE is not yet

the standard of care.� Only protocols with established safety and efficacy

should be used.

Research directions

� Does hypothermia reduce disability in survivors,independent of its effect on mortality?� Does hypothermia change the prognostic implica-

tions of post-hypothermic neurological examina-tion or MR imaging?� Does earlier initiation of cooling, within the first

6 h of life, improve outcome?� Can drug plus hypothermia combinations safely

improve outcome?� Could longer duration or lower target temperature

safely improve outcome?

34 J.D.E. Barks

8. Gray J, Geva A, Zheng Z, Zupancic J. CoolSim: using industrialmodeling techniques to examine the impact of selective headcooling in a model of perinatal regionalization. Pediatrics, inpress.

9. Gunn AJ, Gunn TR. The ‘pharmacology’ of neuronal rescuewith cerebral hypothermia. Early Hum Dev 1998;53(1):19e35.

10. Eicher DJ, Wagner CL, Katikaneni LP, et al. Moderate hypother-mia in neonatal encephalopathy: safety outcomes. PediatrNeurol 2005;32(1):18e24.

11. Leviton A,NelsonKB.Problems withdefinitionsandclassificationsof newborn encephalopathy. Pediatr Neurol 1992;8(2):85e90.

12. Wyatt JS, Gluckman PD, Liu PY, et al. Determinants of out-comes after head cooling for neonatal encephalopathy. Pediat-rics 2007;119(5):912e21.

13. Ambalavanan N, Carlo WA, Shankaran S, et al. Predicting out-comes of neonates diagnosed with hypoxemiceischemic en-cephalopathy. Pediatrics 2006;118(5):2084e93.

14. Liu Y, Barks JD, Xu G, Silverstein FS. Topiramate extends thetherapeutic window for hypothermia-mediated neuroprotec-tion after stroke in neonatal rats. Stroke 2004;35(6):1460e5.

15. Jatana M, Singh I, Singh AK, Jenkins D. Combination of systemichypothermia and N-acetylcysteine attenuates hypoxiceische-mic brain injury in neonatal rats. Pediatr Res 2006;59(5):684e9.

16. Iwata O, Iwata S, Thornton JS, et al. ‘Therapeutic time win-dow’ duration decreases with increasing severity of cerebralhypoxiaeischaemia under normothermia and delayed hypo-thermia in newborn piglets. Brain Res 2007;1154:173e80.

17. InderTE, Hunt RW, MorleyCJ, et al. Randomized trial of systemichypothermia selectively protects the cortexon MRI in term hypo-xiceischemic encephalopathy. J Pediatr 2004;145(6):835e7.

18. RutherfordMA, Azzopardi D, WhitelawA, et al. Mild hypothermiaand the distribution of cerebral lesions in neonates with hypo-xiceischemic encephalopathy. Pediatrics 2005;116(4):1001e6.