Induced hypothermia by central venous infusion: Saline ice slurryversus chilled saline

Terry L. Vanden Hoek, MD; Kenneth E. Kasza, PhD; David G. Beiser, MD, MS;Benjamin S. Abella, MD, MPhil; Jeffery E. Franklin, AAS; John J. Oras, PhD; Jason P. Alvarado, BA;Travis Anderson, BS; Hyunjin Son, BS; Craig L. Wardrip, DVM; Danhong Zhao, MD, PhD;Huashan Wang, MD, MS; Lance B. Becker, MD

Cardiac arrest survivors oftenhave severe neurologic se-quelae that contribute to theirearly mortality and poor func-

tional recovery. Induced postresuscitativehypothermia was first reported almost 40

yrs ago as a technique for improvingfunctional outcomes after cardiac arrest(1, 2). The conceptual appeal of suchtherapeutic hypothermia stemmed pri-marily from the notion that decreases incore temperature lead to decreased cellu-lar metabolic demand. Cerebral meta-bolic demand decreases by approximately6% per degree Celsius (3). Despite this,postresuscitative hypothermia waslargely abandoned as a difficult and re-source-intensive technique with uncer-tain clinical benefits.

However, the clinical induction of hy-pothermia has found widespread accep-tance in the perioperative setting duringcardiac bypass and certain neurosurgicalprocedures. In the surgical setting, un-like cardiac arrest, hypothermia is in-duced before the ischemic insult occurs.Indeed, in the relatively controlled envi-ronment of the operating suite, hypo-thermia induction has been a viable ther-apeutic option that is protective against

conditions of altered blood flow. In addi-tion, hypothermia has also been appliedto the treatment of malignant hyperther-mia, traumatic brain injury, and acutemyocardial infarction (4–7).

Two recent multiple-center, prospec-tive studies of comatose survivors of out-of-hospital cardiac arrest demonstratedimproved neurologic outcome at hospitaldischarge (8) and improved neurologicoutcome and mortality rate at 6 months(9) in patients receiving 12–24 hrs ofhypothermia (32–34°C) after return ofspontaneous circulation. Based on thestrength of these findings, the Interna-tional Liaison Committee on Resuscita-tion recommended the induction of ther-apeutic hypothermia for 12–24 hrs inselected comatose survivors after out-of-hospital cardiac arrest with initial ven-tricular fibrillation (10). The InternationalLiaison Committee on Resuscitation advi-sory statement calls for the developmentof hypothermia-induction techniques

From the University of Chicago, Section of Emer-gency Medicine, Chicago, IL (TLVH, DGB, BSA, JA, TA,LBB); Argonne National Laboratory, Energy TechnologyDivision, Argonne, IL (KEK, JEF, JJO, HS); and theUniversity of Chicago, Animal Resource Center, Chi-cago, IL (CLW).

Supported, in part, by NIH/NHLBI through 1RO1(HL67630) and through a grant from the Laerdal Foun-dation.

Terry Vanden Hoek and Ken Kasza share firstauthorship of this article on the basis of their majorcontributions in working with Dr. Becker in the con-ception, planning, and execution of the experiments;David Beiser is second author for taking primary re-sponsibility for writing the manuscript, analyzing data,and modeling the heat transfer.

Copyright © 2004 by the Society of Critical CareMedicine and Lippincott Williams & Wilkins

DOI: 10.1097/01.CCM.0000134259.59793.B8

Objective: Surface cooling improves outcome in selected co-matose survivors of cardiac arrest. Internal cooling with consid-erable volumes of intravenous cold saline may accelerate hypo-thermia induction. This study compares core temperatures inswine after central catheter infusions of saline ice slurry (salinewith smoothed 100-�m-size ice particles) vs. an equal volume ofchilled saline. We hypothesized that slurry would achieve corehypothermia (32–34°C) more consistently and at a faster rate.

Design: A total of 11 swine were randomized to receive mi-croparticulate ice slurry, chilled saline infusion, or anesthesiaalone in a monitored laboratory setting.

Interventions: Intravenous bolus (50 mL/kg) of slurry or chilled1.5% NaCl saline. Slurry was composed of a 1:1 mixture of ice anddistilled H2O plus NaCl.

Measurements: Cerebral cortex, tympanic membrane, inferiorvena cava, rectal temperatures, electrocardiogram, arterial bloodpressure, and arterial oxygen saturation were recorded for 1 hrafter bolus.

Main Results: Compared with anesthetized controls, core braintemperatures of the saline and slurry groups dropped by 3.4 �0.4°C and 5.3 � 0.7°C (p � .009), respectively. With an infusionrate of 120 mL/min, cooling rates for the saline and slurry groupswere �11.6 � 1.8°C/hr and �18.2 � 2.9°C/hr, respectively,during the first 20 mins. Four of four animals in the slurry groupvs. zero of four animals in the saline group achieved targetcortical temperatures of <34°C.

Conclusions: Cold intravenous fluids rapidly induce hypothermia inswine with intact circulation. A two-phase (liquid plus ice) saline slurrycools more rapidly than an equal volume of cold saline at 0°C. Ice-slurrycould be a significant improvement over other cooling methods whenrate of cooling and limited infusion volumes are important to the clini-cian. (Crit Care Med 2004; 32[Suppl.]:S425–S431)

KEY WORDS: induced hypothermia; therapeutic hypothermia;cooling; resuscitation; cardiac arrest; swine; slurry; intravenous;saline; core temperature

S425Crit Care Med 2004 Vol. 32, No. 9 (Suppl.)

that are rapid, convenient, and minimallyinvasive.

Hypothermia InductionMethodologies: Surface andInternal Cooling

Each method for inducing hypother-mia represents a different compromiseamong invasiveness, cooling rate, tem-perature control, and ease of use. Mini-mally invasive surface cooling techniquessuch as ice packs, forced-air cooling mat-tresses and blankets, and cooling capshave been used in clinical studies of pa-tients with cardiac arrest (8, 9, 11, 12). Inthe Hypothermia After Cardiac Arreststudy, the median time from return ofspontaneous circulation to target coretemperature (32–34°C) was 8 hrs (inter-quartile range, 4–16 hrs), with typicalcooling rates of �1°C/hr (9). Some sur-face methods used in these studies re-quire minimal equipment (e.g., ice packs)and can be initiated by paramedics. Oth-ers, such as cooling mattresses/blankets,require specialized equipment.

Internal cooling techniques such asendovascular cooling catheters have beendeveloped to address the induction,maintenance, and rewarming phases ofhypothermia. Such catheters have dem-onstrated rapid cooling rates in human-sized swine (4.5°C/hr) and nonhumanprimates (6.3°C/hr) (13, 14). Recently,the Cool-MI study utilized endovascularcooling catheters for inducing hypother-mia in awake, but sedated, patients in thesetting of acute ST-segment elevationmyocardial infarction (4). Although moreinvasive, this approach provides fastercooling rates and more precise tempera-ture control than surface cooling. Otherinternal cooling techniques reported in-clude the use of hemodialysis (15). Internalcooling techniques provide faster cooling;however, some of this advantage is lost be-cause additional time is required for theset-up and placement of catheters and sup-port equipment before cooling is initiated.

Recent studies have demonstratedthat an intravenous bolus of chilled sin-gle-phase crystalloid fluid can quickly in-duce hypothermic core temperatures(16–19). Infusion of a large-volume bolus(40 mL/kg) of chilled saline via a femoralcatheter produced an average 2.5°C de-crease in core temperature within ap-proximately 30 mins (17). One study ofcomatose out-of-hospital cardiac arrestsurvivors demonstrated an average 1.7°Cdecrease in bladder temperature after a

bolus (30 mL/kg) of chilled (4°C) lactatedRinger’s (given via peripheral or via cen-tral catheter). In addition, the groupdemonstrated a significant increase inmean arterial pressure. Despite the largebolus volume, none of the postarrest pa-tients developed clinical or radiographicevidence of pulmonary edema (16).

The cooling rates achieved by intrave-nous chilled saline seem quite rapid, yet thelarge bolus volumes required to reach hy-pothermic temperatures remain a potentialconcern given the variation in cardiopul-monary dysfunction that may occur afterreturn of spontaneous circulation. In addi-tion to cardiac arrest, other resuscitationsettings may require high cooling potentialwith limited fluid volumes. For example,recent strategies proposed for the treat-ment of hemorrhagic shock or traumaticbrain injury include the use of hypertonicsaline or the induction of mild hypother-mia. Both strategies may attenuate inflam-mation associated with global ischemia/reperfusion (7, 20). However, highersodium loads and concerns of diluting clot-

ting factors or increasing edema limit thevolume of saline that can be given.

Development of a Saline Slurryfor Intravenous Application

We tested a microparticulate slurry(MPS) coolant originally developed by Ar-gonne National Laboratory for industrialcooling applications. MPS, a two-phasefluid composed of smooth globular iceparticles of �100 �m in diameter sus-pended in a liquid carrier medium (Fig.1), has an inherent thermodynamic ad-vantage over liquid saline, stemmingfrom the energy required to melt ice insolution. For ice solutions, this so-calledlatent heat of fusion is approximately 80cal/g solution. Accordingly, it requires 74kcal of energy to melt and warm a1000-mL sample of slurry composed of a1:1 mixture of ice and saline from 0 to34°C. By contrast, warming a similarsample of saline requires only 34 kcal.

A proprietary procedure for chemi-cally and thermally smoothing individual

Figure 1. Microparticulate slurry (left, top and bottom) is composed of smooth globular (100 �m) iceparticles with superior fluid dynamic properties that make it suitable for pumping through intrave-nous catheters. Dendritic ice (right, top and bottom), as produced by commercial slush beverage–making machines, cannot be readily pumped, and produces plugging when poured.

S426 Crit Care Med 2004 Vol. 32, No. 9 (Suppl.)

ice particles is used to create a highlyflowable MPS that can pass throughsmall-diameter tubing without plugging(21). By contrast, ordinary ice-liquid sus-pensions (e.g., slush beverages) are com-posed of dendritic ice crystals that do notflow at high ice-particle loading valueswithout plugging. The prototype MPSyields saline with a salt concentration of1.5% by weight when fully melted. Al-though this MPS is necessarily hyper-tonic, future design improvements willlikely yield solutions with decreased saltconcentrations.

METHOD

We hypothesized that an intravenous bolusof MPS would induce hypothermia (32–34°C)more quickly than an equal bolus of chilledsaline. We studied 11 farm swine, Sus scrofadomestica, using a protocol approved by theInstitutional Animal Care and Use Committee.The swine were anesthetized with an intra-muscular cocktail of ketamine, xylazine, tilet-amine, and zolazepam (each at 2.2 mg/kg),followed by 1–2% isoflurane inhalant formaintenance of an appropriate depth of anes-thesia during mechanical ventilation. A mod-ified 9-Fr catheter was then placed in the rightfemoral vein for delivering the cooling solu-tions. Animals underwent bilateral temporalcraniotomies for placement of thermocouplesat a depth of 1.25 cm for recording corticaltemperatures. Thermocouple probes, embed-ded in a thermoconductive paste, were in-serted into the ear canal to the tympanicmembrane for recording bilateral tympanicmembrane temperatures. Rectal and inferiorvena cava (IVC) temperatures were also re-corded. Temperature data were acquired at1-min intervals.

Three groups of animals were studied: sa-line infusion, MPS infusion, and anesthetizedcontrols. For the saline group, a bolus (50mL/kg target dose) of chilled (0–1°C) saline(1.5% NaCl by weight) was pumped throughinsulated tubing using a roller pump at 120mL/min. During administration, temperatureswere recorded at 1-min intervals for 60 minsafter the start of the infusion. The animalswere then killed. The MPS group received thesame volume (50 mL/kg) of the MPS (equilib-rium temperature of �1 to 0°C) initially, com-posed of a 1:1 mixture of distilled ice and H2Oplus 1.5% non-iodinated NaCl by weight. Al-though the initial ice loading of this mixture is50%, a substantial amount of melting occursduring slurry processing to yield a deliveredvalue of 15–20% ice by weight. Control datawere obtained from animals utilizing the sameanesthesia protocol and instrumentation with-out cooling.

RESULTS

The experimental group characteris-tics are detailed in Table 1. The weights ofthe saline and MPS animals were statis-tically indistinguishable. The averageweight of the control group was statisti-cally less than the saline group. Giventheir smaller thermal mass, the controlgroup would be expected to display in-creased cooling rates after induction ofgeneral anesthesia, thus overestimatingthe effect of sedation alone on coolingrate (22). The delivered weight-based dos-ages of intravenous coolants, initial cor-tical temperatures, infusion time periods,mean arterial pressure, and heart rate werestatistically indistinguishable in the salineand MPS groups. All experiments were per-formed in the same climate-controlled an-imal surgical suite set for 21–23°C.

With cooling, brain cortex tempera-tures in the saline and MPS groups de-

creased rapidly to their minima duringthe average infusion period of approxi-mately 18 mins (Fig. 1). Minimum corti-cal temperatures for the MPS group weresignificantly lower than those of thechilled saline group (Fig. 2, Table 2). Thecortex in each of the MPS swine reachedtemperatures of �34°C within an averagetime period of 7.0 (range, 3–10) mins,with an average absolute change in tem-perature of 5.3 � 0.7°C, when the 20%ice-loaded MPS was delivered at a rate of120 mL/min. The saline pigs achieved a3°C temperature decrease after an aver-age of 16.3 (range, 13–19) mins whennear-0°C chilled saline was delivered at120 mL/min; however, due to average ini-tial core temperatures of �37°C, nonecrossed into the target range (32–34°C).

Average initial cooling rates using aninfusion rate of 120 mL/min were esti-mated by averaging linear least-squares

Figure 2. Average cortical temperatures with injection of saline and microparticulate slurry (MPS).Data presented as mean � SD.

Table 1. Experimental group characteristics

Characteristic Control Saline MPS p Valuea

n 3 4 4Weight, kg 39.0 � 1.7 45.4 � 2.0 42.2 � 2.8 .12Dose, mL/kg NA 49.7 � 0.7 50.4 � 0.8 .25Infusion time, min NA 18.8 � 0.9 17.7 � 1.0 .16MAP, mm Hg 83.8 � 12 78.3 � 22 77.2 � 15 .94HR, beats/min 109 � 28 94.5 � 11 83.0 � 19 .35

MPS, microparticulate slurry; NA, not applicable; MAP, mean arterial pressure; HR, heart rate.ap values for t-test comparing saline and MPS groups only; weights of the control and MPS groups

were statistically different (p � .005). Data are presented as mean � SD.

S427Crit Care Med 2004 Vol. 32, No. 9 (Suppl.)

approximations (R � .95) of the corticaltemperature data sets between the initialand minimum temperature values (Table2). Because the control pigs did notachieve minimum temperatures, theirinitial rates were calculated by choosingthe temperature at the end of the averageinfusion period. At the same rate of infu-sion, the average initial cooling rate forthe MPS group was approximately 50%faster than the saline group. After theinfusion period, cortical temperaturespartially rebounded to temperatures thatwere still significantly below the mini-mum cortical temperature of the controlgroup.

Average rectal, tympanic membrane,cortical, and central venous temperaturesfor the saline group are presented in Fig-ure 3. As expected, the venous tempera-ture (from the thermocouple placed inthe IVC) displayed the largest reductionin temperature during the infusion pe-riod. Like the cortical temperatures, theIVC and tympanic membrane tempera-tures reached minima followed by partialrebounds after the cessation of infusion.By contrast, the rectal temperature didnot reach a minimum and continued todecrease after infusion (although at aslower rate). After the infusion, the tem-peratures equilibrated with each other af-ter approximately 40 mins.

MPS induced a more profound instan-taneous decrease in IVC temperature be-fore dramatically rebounding to equili-brate with the cortical and tympanicmembrane temperatures (Fig. 4). Asnoted for the cortical temperaturesabove, the rectal, tympanic membrane,and IVC temperatures did not continue todrop after rebound. Average tympanicmembrane temperatures followed cool-ing trajectories that were qualitativelysimilar to cortical temperatures, reach-ing statistically equivalent minima (p �.90 for paired t-test) with a 2- to 4-minlag between cortical and tympanic ther-mocouple readings.

Finally, heart rate, mean arterial pres-sure, and arterial oxygen saturation re-mained statistically constant throughoutthe experimental period. In addition, noqualitative changes in the electrocardio-gram wave morphologies were noted.

DISCUSSION

The induction of hypothermia has re-cently been recommended for selectedcomatose cardiac arrest patients after re-turn of spontaneous circulation (8–10).

Given this international advisory state-ment and its clinical basis, there is a clearneed for the development of hypother-mia-induction techniques that are rapid,safe, and simple to implement in theemergent setting. Given concerns aboutfluid volume in resuscitation settings(both for cardiac arrest and hemorrhagicshock), coolants with greater cooling ca-pacities, such as microparticulate iceslurries, may prove useful in this setting.

Characteristics of Intravenous Cool-ing. Several studies have sought to un-derstand the mechanisms of human ther-moregulation through theoretical modelsand experimental measures (17, 22–25).

The simplest models partition the bodyinto core and peripheral compartments.The core compartment, which comprisesthe highly perfused thoracoabdominal or-gans and brain, remains at a relativelyuniform temperature, despite constantenvironmental changes. In our study,cortical brain temperatures served as thegold standard for recording core temper-ature; however, we found good correla-tion with tympanic membrane tempera-tures with a 2- to 4-min time lag. Aspreviously observed, we found that rectaltemperature was not a good proxy forcore temperature (26). The peripheralcompartment, which includes the limbs

Figure 3. Average rectal, tympanic membrane, cortical, and central venous temperatures during coldsaline infusion. Inferior vena cava temperature reached 34.0 � 0.3°C at 17 mins.

Table 2. Experimental cortical temperatures and cooling rates

Variable Control Saline MPS p Valuea

Temperatures, °CInitial 37.5 � 0.3 38.0 � 0.4 36.7 � 1.3 .13Minimum 37.3 � 0.3c 34.5 � 0.4 31.4 � 0.9 .002Absolute �T 0.2 � 0.1 3.4 � 0.4 5.3 � 0.7 .009Reboundb 36.8 � 0.5c 35.3 � 0.4 32.8 � 1.3 .028

Time to, minsT � 34°C NA �60 7.0 � 2.9 NA�T � 3°C NA 16.3 � 2.8 8.3 � 1.0 �.002

Cooling rate, °C/hrInitial 18 mins �0.89 � 0.7 �11.6 � 1.8 �18.2 � 2.9 �.002Final 20 mins �1.62 � 0.35 �0.89 � 0.46 �0.18 � 0.40 .07

MPS, microparticle slurry; �T, change in temperature; T, temperature; NA, not applicable.ap values compare saline and MPS groups; brebound temperatures defined as maximum temper-

ature after the infusion; ccontrol minimum and rebound defined as temperature at the end of theaverage infusion period and at 40 mins, respectively. Data are presented as mean � SD. Initial (18 mins)and final (20 mins) cooling rates were calculated by averaging least-squares linear approximations ofindividual data sets.

S428 Crit Care Med 2004 Vol. 32, No. 9 (Suppl.)

and skin, is characterized by a labile tem-perature that compensates for metabolicand environmental changes.

Temperatures in the anesthetized con-trol animals decreased at absolute ratesbetween 0.9 and 1.6°C/hr, which are con-sistent with temperature changes notedduring the first hour after the inductionof general anesthesia in humans. Thisdecrease is primarily the result of redis-tribution of heat from the core to theperiphery rather than environmental lossor decreased metabolic rate (27).

Our data suggest that a large resusci-tation volume (50 mL/kg) intravenousbolus of MPS can rapidly induce mild tomoderate hypothermia in the core com-partment of swine with intact circulation.A bolus of chilled saline did not achievecortical temperatures of �34°C but didproduce a significant mean temperaturedecrease (3.4°C) slightly greater thanprevious studies utilizing smaller doses(30–40 mL/kg) of chilled saline in hu-man volunteers and patients with cardiacarrest (16, 17).

Unlike surface cooling, intravenouscoolants induce extremely rapid corticalcooling rates. In the current study, cor-tical temperature minima were reachedwithin 1 min after completion of the 50mL/kg bolus. This result suggests thatcooling rate is heavily dependent on therate of infusion such that increased ratesof administration will likely yield fastercooling rates.

After the intravenous bolus, core tem-peratures uniformly displayed a pro-nounced partial temperature rebound(Fig. 1). Figure 4 suggests that the ob-served rebound could be secondary to re-distribution of heat from relativelywarmer areas such as the rectum, ab-dominal vessels, or peripheral tissues. Ra-jek et al. (17) noted a similar reboundeffect in their detailed study of intrave-nous cooling in healthy volunteer sub-jects. By contrast, they concluded thatthe rebound was a result of metabolicheat constrained to the core via periph-eral vasoconstriction. Despite this partialrebound, the MPS group core tempera-tures remained at �34°C.

Changes in Heat Content. As thermo-dynamic theory predicts, two-phase MPSinduces lower core temperatures thanchilled saline. To quantify this thermaladvantage, we calculated the amount ofheat absorbed by the coolants duringeach trial. The heat content of a liquidcoolant increases as it absorbs energyfrom the body (Appendix, Eq. 1).

�Q � m � C � �T [1]

Heat content changes for the saline cool-ant were calculated based on the mini-mum and rebound temperatures (Table3). Heat content changes for the MPScoolant were calculated using an esti-mated ice load of 20% by weight (Appen-dix, Eq. 2).

�QMPS � mMPS � C � �T � %ice � mMPS � L

[2]

We calculated the energy dissipated bythe core compartment by assuming anaverage tissue-specific heat of 0.83kcal·kg�1·°C�1 and an approximate 60/40partition between the mass of the coreand peripheral compartments for hu-

mans (27). Decreases in core compart-ment heat content for the control, saline,and MPS pig groups were calculated atthe minimum and rebound temperatures(Table 3).

When calculated at the minimumtemperature, total increases in coolantheat content are similar to the corre-sponding decreases in core heat content(Table 3). However, if we assume thatheat absorption is proportionally distrib-uted throughout the peripheral and corecompartments, the proportional amountof coolant heating attributable to the coreis only 60% of the total value (parenthet-ical values in Table 3). Once adjusted, thecore heat loss exceeds the amount pro-portionally absorbed by the coolant. Such

Figure 4. Average rectal, tympanic membrane, cortical, and central venous temperatures duringmicroparticulate slurry infusion. Note that the central venous temperature reached 28.6 � 1.1°C.

Table 3. Change in core heat content

Change in Heat Content, kcal Controla Saline MPS

At minimum temperaturePig core �3.9 � 2.1 �77.2 � 6.7 �110 � 10.3

Coolant NA 80.1 � 3.4 103 � 6.7(Attributed to core)b NA (48) (61)Excess core cooling 3.9 29 49

At rebound temperaturePig core �14.7 � 5.5 �61.0 � 1.5 �81.5 � 2.4

Coolant total NA 81.8 � 3.2 106 � 6.7(Attributed to core)b NA (49) (64)Excess core cooling 14.7 12 17.5

MPS, microparticle slurry; NA, not applicable.aControl group heat fluxes calculated using temperatures at 18 (minimum) and 40 (rebound) mins,

respectively; bcoolant heat content change attributed to core (in parentheses) was calculated bymultiplying the coolant total by the 60% core partition estimate. Data are presented as mean � SD.

S429Crit Care Med 2004 Vol. 32, No. 9 (Suppl.)

“excess core cooling,” previously de-scribed in a study of human intravenouscooling, suggests that the coolant is pref-erentially constrained to the core com-partment (17). Alternatively, this dispar-ity may reflect inaccuracies in our 60/40compartment partition assumption.

Risks and Benefits of IntravenousCooling. Clinical studies confirm that hu-mans are difficult to cool. The adult bodycombats accidental hypothermia by in-creasing metabolic heat productionthrough shivering and by decreasing en-vironmental losses by thermally isolatingthe core from the peripheral compart-ment via peripheral vasoconstriction (17,22, 28). These compensatory mechanismsare attenuated during induction of gen-eral anesthesia, thus facilitating corecooling with surface methods (27, 29).However, in the postarrest setting, ther-mal isolation between the peripheral andcore compartments is likely increased bythe overall “adrenergic” state of the post-arrest patient after administration of va-soactive medications. Our results suggestthat intravenous coolants, when admin-istered through a central catheter, maybypass the peripheral compartment com-pletely. Although intravenous coolantsmay bypass the thermal resistance of theperipheral compartment, this thermalshunting may be attenuated when theyare administered through a peripheralvein.

The heat content equation (Appendix,Eq. 2) shows that the cooling capacity ofMPS is determined, in part, by the totaldosage and ice-loading percentage. Wecan apply this relationship to the averagedata from the saline and MPS groups toestimate a range of dosage/ice-load com-binations that would induce similar cor-tical cooling results (Fig. 5). The topcurve, which is based on the MPS data,extrapolates the dosage/ice-load combi-nations that would produce minimumcore temperatures approximately equal tothe MPS group. The bottom curve com-

binations would lead to cooling resultssimilar to the saline group. The regionbetween the curves represents combina-tions that would lead to minimum tem-peratures between the MPS and salinegroup minima. Recall that the salinegroup did not quite reach the targetrange (32–34°C), whereas the MPS groupproduced minimum temperatures of�32°C. From the top curve of Figure 5,we can estimate that it would require abolus of approximately 75 mL/kg of saline(i.e., 0 % ice load) to produce coolingequivalent to 50 mL/kg of 20% ice MPS.Similarly, it would take a bolus of approx-imately 34 mL/kg of 20% ice MPS toreproduce the saline results 50 mL/kgsaline. As illustrated in Figure 5, whenMPS ice loads are increased to �20%, thebolus volumes required for a given tem-perature change decrease significantly.With more highly loaded MPS, we expectsmaller volumes to achieve the same de-gree of cooling. Minimizing the total vol-ume of an intravenous coolant bolusmight be important for avoiding pulmo-nary edema after cardiac arrest and he-modilution of clotting factors after hem-orrhagic shock. Supporting this goal ofreducing coolant volume, our collabora-tors at Argonne National Laboratory haverecently improved MPS ice loading to lev-els up to 50% by weight. From Figure 5,we can predict that a 23 mL/kg bolus of50% ice-load MPS would produce tem-

perature minima similar to those ob-served in the saline group.

Another important consideration iswhether the cooled blood traversing theheart may produce adverse effects. Withthe MPS group, the IVC reached a mini-mum temperature of 28.6 � 1.1°C, whichif transmitted to the myocardium, couldtheoretically induce ventricular fibrilla-tion. Although we saw no adverse effectson hemodynamics or electrocardiogramtracings, this study was not by designpowered to detect low-frequency events.A slower infusion rate or smaller boluswould likely attenuate the IVC tempera-ture drop if required. The IVC tempera-ture was almost 5°C lower than the rectaltemperature for the MPS group and 2°Cin the saline group. This reflects a ther-mal compartmentalization that could beeven more pronounced during the low-flow setting of CPR.

In conclusion, cold intravenous fluidscan rapidly induce hypothermia in pigswith intact circulation. A two-phase (liq-uid plus ice) MPS cools more rapidly andeffectively than cold saline at 0°C. Iceslurry could be a significant improve-ment over other cooling methods whenspeeds of cooling and low infused vol-umes are important to the clinician. Fur-ther work on methods for the intrave-nous induction of hypothermia iswarranted.

Figure 5. Theoretical dosage vs. ice load extrapolated from saline (bottom curve) and microparticulateslurry (top curve) cortical cooling minima data points (indicated by circles). Microparticulate slurrydosage/ice-load combinations falling between the curves will produce core temperature minimumswithin a range of 31.4 � 0.9°C to 34.5 � 0.4°C.

T wo-phase (liquid

plus ice) saline

slurry cools more

rapidly than an equal vol-

ume of cold saline at 0°C.

S430 Crit Care Med 2004 Vol. 32, No. 9 (Suppl.)

ACKNOWLEDGMENTS

We thank the staff of the University ofChicago Animal Resources Center andLynne Harnish for her administrativesupport.

REFERENCES

1. Williams G, Spencer F: The clinical use ofhypothermia following cardiac arrest. AnnSurg 1958; 148:462–468

2. Benson D, Williams G, Spencer F, et al: Theuse of hypothermia after cardiac arrest.Anesth Analg 1959; 38:423–428

3. Steen PA, Newberg L, Milde JH, et al: Hypo-thermia and barbiturates: Individual andcombined effects on canine cerebral oxygenconsumption. Anesthesiology 1983; 58:527–532

4. Dixon SR, Whitbourn RJ, Dae MW, et al:Induction of mild systemic hypothermia withendovascular cooling during primary percu-taneous coronary intervention for acutemyocardial infarction. J Am Coll Cardiol2002; 40:1928–1934

5. Clifton GL, Miller ER, Choi SC, et al: Lack ofeffect of induction of hypothermia after acutebrain injury. N Engl J Med 2001; 344:556–563

6. Einer-Jensen N, Khorooshi MH, PetersenMB, et al: Rapid brain cooling in intubatedpigs through nasal flushing with oxygen:Prevention of brain hyperthermia. Acta VetScand 2001; 42:459–464

7. Bayir H, Clark RS, Kochanek PM: Promisingstrategies to minimize secondary brain in-jury after head trauma. Crit Care Med 2003;31(1 Suppl):S112–S117

8. Bernard SA, Gray TW, Buist MD, et al: Treat-ment of comatose survivors of out-of-hospital cardiac arrest with induced hypo-thermia. N Engl J Med 2002; 346:557–563

9. The Hypothermia after Cardiac Arrest StudyGroup: Mild therapeutic hypothermia to im-prove the neurologic outcome after cardiacarrest. N Engl J Med 2002; 346:549–556

10. Nolan JP, Morley PT, Vanden Hoek TL, et al:Therapeutic hypothermia after cardiac ar-rest: An advisory statement by the AdvancedLife Support Task Force of the InternationalLiaison Committee on Resuscitation. Circu-lation 2003; 108:118–121

11. Callaway CW, Tadler SC, Katz LM, et al:Feasibility of external cranial cooling duringout-of-hospital cardiac arrest. Resuscitation2002; 52:159–165

12. Hachimi-Idrissi S, Corne L, Ebinger G, et al:Mild hypothermia induced by a helmet de-vice: A clinical feasibility study. Resuscita-tion 2001; 51:275–281

13. Dae MW, Gao DW, Ursell PC, et al: Safety andefficacy of endovascular cooling and rewarm-ing for induction and reversal of hypother-mia in human-sized pigs. Stroke 2003; 34:734–738

14. Mack WJ, Huang J, Winfree C, et al: Ultra-rapid, convection-enhanced intravascularhypothermia: A feasibility study in nonhu-man primate stroke. Stroke 2003; 34:1994–1999

15. Laurent I, Adrie C, Vinsonneau C, et al: High-volume hemofiltration to improve prognosisafter cardiac arrest. Abstr. Circulation 2003;108(Suppl):1496

16. Bernard S, Buist M, Monteiro O, et al: In-duced hypothermia using large volume, ice-cold intravenous fluid in comatose survivorsof out-of-hospital cardiac arrest: A prelimi-nary report. Resuscitation 2003; 56:9–13

17. Rajek A, Greif R, Sessler DI, et al: Core cool-ing by central venous infusion of ice-cold(4°C and 20°C) fluid: Isolation of core andperipheral thermal compartments. Anesthe-siology 2000; 93:629–637

18. Frank SM, Raja SN, Bulcao C, et al: Age-related thermoregulatory differences duringcore cooling in humans. Am J Physiol RegulIntegr Comp Physiol 2000; 279:R349–R354

19. Baumgardner JE, Baranov D, Smith DS, etal: The effectiveness of rapidly infused intra-venous fluids for inducing moderate hypo-thermia in neurosurgical patients. AnesthAnalg 1999; 89:163–169

20. Takasu A, Stezoski SW, Stezoski J, et al: Mildor moderate hypothermia, but not increasedoxygen breathing, increases long-term sur-vival after uncontrolled hemorrhagic shockin rats. Crit Care Med 2000; 28:2465–2474

21. Hayashi K, Kasza KE: Ice slurry cooling re-search: Effects of microscale ice particlecharacteristics and freezing-point depressantadditives on ice slurry fluidity. Am Soc Heat-ing Refrigeration Air-Conditioning EngTrans 2001; 107:346–351

22. Sessler DI: Perioperative heat balance. Anes-thesiology 2000; 92:578–596

23. Cornew RW, Houk JC, Stark L: Fine controlin human temperature regulation system.J Theor Biol 1967; 16:406–426

24. Sessler DI: Mild perioperative hypothermia.N Engl J Med 1997; 336:1730–1737

25. Sessler DI, Sessler AM: Experimental deter-mination of heat flow parameters during in-duction of general anesthesia. Anesthesiol-ogy 1998; 89:657–665

26. Stone JG, Young WL, Smith CR, et al: Dostandard monitoring sites reflect true braintemperature when profound hypothermia israpidly induced and reversed? Anesthesiol-ogy 1995; 82:344–351

27. Matsukawa T, Sessler DI, Sessler AM, et al:Heat flow and distribution during inductionof general anesthesia. Anesthesiology 1995;82:662–673

28. Kurz A, Sessler DI, Birnbauer F, et al: Ther-moregulatory vasoconstriction impairs ac-tive core cooling. Anesthesiology 1995; 82:870–876

29. Plattner O, Xiong J, Sessler DI, et al: Rapidcore-to-peripheral tissue heat transfer dur-ing cutaneous cooling. Anesth Analg 1996;82:925–930

APPENDIX

Calculating Heat Content. The heatcontent of a coolant increases as it ab-sorbs energy from the body. For a single-phase coolant such as saline, this in-crease in heat content is directly relatedto the change in coolant temperature(Eq. 1) through a constant known as thespecific heat (1 kcal·kg�1·°C�1 for saline).

The net amount of energy absorbed bythe saline during the initial cooling pe-riod is exactly determined by the initialand minimum coolant temperatures.Using an aggregate value for the spe-cific heat of human tissue (0.83kcal·kg�1·°C�1) and a core compart-ment of approximately 60% of the body,we used Equation 1 to calculate an es-timated change in the core compart-ment’s heat content (Table 3) for thecontrol, saline, and MPS groups (27).

Where �Q � heat flux (kcal), m �saline coolant (or core) mass (kg), C �specific heat of coolant (or core tissue,kcal·kg�1·°C�1), and �T � temperaturechange of coolant (or core, °C).

For the MPS coolant, a term is addedto account for the energy required tomelt the ice particles during warming(Eq. 2). The energy required to promotethis phase change is determined by thelatent heat of fusion and mass of ice insolution. Ice particles initially comprise50% of the MPS mass, although this frac-tion decreases to approximately 15–20%as a result of the smoothing process. Forsimplicity, we ignore the negligibleamount of energy required to warmslurry ice from �1°C to its 0°C meltingpoint.

Where �QMPS � change in MPS heatcontent (kcal), mMPS � MPS mass (kg),%ice � percentage of ice loading (15–20%), C � specific heat of saline (1kcal·kg�1·°C�1), L � latent heat of fusionfor ice (80 kcal/kg), and �T � changeMPS temperature (°C).

Equation 2 can be rearranged to illus-trate the inverse relationship betweenMPS dosage and percentage of ice load(Eq. 3).

mMPS ��QMPS

C � �T � %ice � L[3]

We can extrapolate combinations of MPSmass and ice load that would producetemperature minima observed in the sa-line and MPS groups (Fig. 5).

S431Crit Care Med 2004 Vol. 32, No. 9 (Suppl.)