comparison of endovascular and surface cooling during unruptured cerebral aneurysm repair: the tcas...

CLINICAL STUDIES

COMPARISON OF ENDOVASCULAR

AND SURFACE COOLING DURING UNRUPTURED

CEREBRAL ANEURYSM REPAIR

Gary K. Steinberg, M.D.,Ph.D.Stanford University MedicalCenter, Stanford, California

Christopher S. Ogilvy, M.D.Massachusetts General Hospital,Harvard University,Boston, Massachusetts

Lawrence M. Shuer, M.D.Stanford University MedicalCenter, Stanford, California

E. Sander Connolly, Jr., M.D.Columbia-Presbyterian MedicalCenter, New York, New York

Robert A. Solomon, M.D.Columbia-Presbyterian MedicalCenter, New York, New York

Arthur Lam, M.D.Harborview Medical Center,University of Washington,Seattle, Washington

Neal F. Kassell, M.D.University of Virginia MedicalCenter, Charlottesville, Virginia

Christopher J. Baker, M.D.Florida Hospital, Orlando, Florida

Steven L. Giannotta, M.D.LAC/USC Medical Center, Universityof Southern California UniversityHospital, Los Angeles, California

Kevin M. Cockroft, M.D.Milton S. Hershey Medical Center,Pennsylvania State University,Hershey, Pennsylvania

Teresa E. Bell-Stephens, R.N.Stanford University MedicalCenter, Stanford, California

Robin L. Allgren, M.D.,Ph.D.Innercool Therapies,San Diego, California

Reprint requests:Gary K. Steinberg, M.D., Ph.D.,Department of Neurosurgery,Stanford University School ofMedicine, 300 Pasteur Drive R200,Stanford, CA 94305-5327.Email: [email protected]

Received, September 16, 2003.

Accepted, March 24, 2004.

OBJECTIVE: To compare endovascular versus surface methods for the induction andreversal of hypothermia during neurosurgery in a multicenter, prospective, random-ized study.METHODS: Patients undergoing elective open craniotomy for repair of an unrupturedcerebral aneurysm (n � 153) were randomly assigned (2:1) to undergo whole-bodyhypothermia to 33°C, either with an endovascular cooling device placed in the inferiorvena cava via the femoral vein (n � 92) or with a surface convective air blanket (n �61). Active rewarming was accomplished using the same devices.RESULTS: Cooling rates in endovascular and surface blanket groups averaged 4.77 and0.87°C/h, respectively (P � 0.001). When the first temporary arterial or aneurysm clipwas placed, 99% of endovascular patients and 20% of surface blanket patients hadreached the target of 33°C (P � 0.001). Obese patients were cooled efficiently with theendovascular approach (3.56°C/h). Rewarming rates averaged 1.88°C/h for endovas-cular patients and 0.69°C/h for surface blanket patients (P � 0.001). By the end ofsurgery, 89 and 53% of these patients, respectively, had rewarmed to at least 35°C (P� 0.001). On leaving the operating room, 14% of endovascular patients and 28% ofsurface blanket patients were still intubated (P � 0.035). The overall safety of the twoprocedures was comparable. No clinically significant catheter-related thrombotic,bleeding, or infectious complications were reported in the endovascular group.CONCLUSION: Endovascular cooling provided superior induction, maintenance, andreversal of hypothermia compared with the surface blanket, without an increase incomplications. Endovascular cooling may have clinical benefit for patients undergoingcerebrovascular surgery, as well as patients with acute stroke, head injury, or acutemyocardial infarction.

KEY WORDS: Aneurysm repair, Clinical trial, Cooling, Hypothermia, Intravascular cooling device,Rewarming

Neurosurgery 55:307-315, 2004 DOI: 10.1227/01.NEU.0000129683.99430.8C www.neurosurgery-online.com

Mild hypothermia has been shown toprotect against experimental cere-bral ischemic damage and to im-

prove neurological outcome after cardiac ar-rest in patients (1, 3, 6, 13–15). Hypothermia isused frequently during neurosurgical proce-dures to optimize neuroprotection during in-traoperative ischemia (13, 16, 18). Althoughmild hypothermia has not been demonstratedto be beneficial during aneurysm surgery in aprospective randomized trial, many cerebro-vascular surgeons in North America use mildhypothermia routinely for all aneurysm sur-

gery, based on the very compelling evidenceof its neuroprotective benefit in experimentalstroke models and their belief that the risks ofmild hypothermia are extremely low. Currentsurface methods for inducing hypothermia in-clude convective air-cooled or water-cooledblankets and pads or the surface application ofice packs or alcohol. Parenteral methods ofhypothermia induction include the intrave-nous infusion of cold saline or cold bladderlavage. All of these methods are cumbersomeand slow to induce desired temperaturechanges. Importantly, they also cannot easily

NEUROSURGERY VOLUME 55 | NUMBER 2 | AUGUST 2004 | 307

provide precise control of temperature at a predefined targettemperature. For example, in one study of patients with isch-emic stroke, surface cooling methods were applied in an at-tempt to induce hypothermia to a target of 32.0°C (5). Evenwith a combination of a cooling blanket and whole-body icewater and alcohol rubs, the target temperature was achievedonly after a mean of 3.5 hours, and 9 (90%) of the 10 patientsstudied exhibited “overshoot” beyond the hypothermia targetof 32.0°C to as low as 28.4°C. Thus, improved methods forinducing and controlling hypothermia are needed.

Endovascular techniques for inducing hypothermia mayhave advantages over surface cooling methods. By inserting acooling catheter directly into a central vessel (such as the venacava), one can directly cool the blood, resulting in much morerapid and controllable cooling of internal organs than can beachieved with surface methods. One type of endovascularcooling system is being tested in acute stroke patients (9). Thepurpose of this study was to compare the safety and thetemperature management performance of an endovascularsystem with that of a standard surface system (the convectiveair blanket) in actual clinical practice among patients under-going aneurysm surgery. This endovascular catheter coolingsystem has been shown to be safe and to provide rapid andreliable hypothermia in pigs (7) and in a primate model ofhemispheric stroke (12).

PATIENTS AND METHODS

The study protocol was approved by each institution’s in-stitutional review board. Written informed consent was ob-tained from each patient. A data safety monitoring boardindependent of the sponsor and study sites reviewed safetyinformation throughout the study to ensure that study partic-ipants were not exposed to unreasonable risk.

Patient Population

Patients aged 18 to 70 years who were scheduled for elec-tive open craniotomy for repair of an unruptured cerebralaneurysm were enrolled. Patients were not eligible if any ofthe following conditions were present: the aneurysm had rup-tured within 30 days before the scheduled surgery date; therewas a contraindication to central venous catheter placementvia femoral vein access; there were contraindications to hypo-thermia, such as known hematological dyscrasias that affectthrombosis (cryoglobulinemia, sickle cell disease, serum coldagglutinins) or vasospastic disorders (such as Raynaud’s orBuerger’s disease); there was a family history of malignanthyperthermia; significant cardiac disease or myocardial infarc-tion had occurred within 90 days of scheduled surgery date;there was severe peripheral vascular disease; the induction ofbarbiturate coma was anticipated; serum creatinine levelswere more than 1.6 mg/dl; or there was a history of heparin-induced thrombocytopenia.

Temperature Management Procedures

All patients were to undergo hypothermia to a target esoph-ageal temperature of 33°C during their neurosurgical proce-dure and then to be actively rewarmed to normothermia (atarget of 36°C) at the end of surgery. Patients were random-ized in a 2:1 ratio to undergo active temperature managementwith either an endovascular cooling system (endovasculargroup) or the institution’s standard commercially availableconvective air blanket system (surface blanket group). Thedevices used in the study are described below.

Endovascular Temperature Management

Endovascular temperature management was performedwith the Celsius Control System (Innercool Therapies, SanDiego, CA) and a 14-French Celsius Control catheter. TheCelsius Control Catheter (Fig. 1A) exchanges heat with thebody by circulating cool (or warm) saline in a closed loopmanner within the catheter, which is placed in the inferiorvena cava via femoral venous access. The catheter has aheparin-coated flexible metallic heat transfer surface that fa-

FIGURE 1. Photographs showing the Celsius Control System intravascu-lar catheter (A) and the temperature control console (B).

STEINBERG ET AL.

308 | VOLUME 55 | NUMBER 2 | AUGUST 2004 www.neurosurgery-online.com

cilitates the extraction of heat from the blood (during coolingmode) or the delivery of heat to the blood (during warmingmode) as blood flows by the catheter. During cooling with theCelsius Control System, cold sterile saline (approximately 2°C)is circulated within the catheter. Thus, the temperature of thecatheter tip during cooling is typically 2 to 4°C. However, theaverage temperature of the blood moving past the cathetersurface is usually within 1°C of the patient’s body temperatureat that time. The catheter is connected to the Celsius ControlSystem console (Fig. 1B), which controls the temperature andflow rate of the fluid that is circulated within the catheter andadjusts them in response to patient temperature feedback(provided by an esophageal temperature probe connected tothe console) as needed to achieve and maintain the targethypothermic or normothermic temperature.

To initiate cooling with the system, the catheter first wasinserted in the operating room after induction of general an-esthesia via a standard introducer sheath into a femoral veinand was advanced to the inferior vena cava until the distal tipof the catheter was just below the right atrium. Correct cath-eter placement was confirmed via portable fluoroscopy or anabdominal x-ray. The desired target temperature (33°C for thisstudy) was set via the console control panel and the coolingmode was activated. As soon as the hypothermic target tem-perature was reached, the Celsius Control System automati-cally began to turn off and on intermittently as needed tomaintain the patient at target temperature. During this main-tenance mode, an ancillary convective air blanket placed overthe patient was turned on to the medium setting (38 � 3°Csurface temperature) to enhance peripheral vasodilatation andthermal equilibrium.

To initiate active rewarming near the end of surgery, thetarget temperature was reset to normothermia via the consolecontrol panel. The Celsius Control System then circulatedwarm saline through the catheter to rewarm the blood as itcirculated near the catheter. The catheter was removed at theend of surgery. As soon as the femoral sheath was removed,direct pressure was applied to the femoral access site untilhemostasis was achieved.

Surface Temperature Management

Surface temperature management was provided using thestandard commercially available surface convective air blan-ket system used at the institution before the study (usuallyeither a Bair Hugger [Augustine Medical, Eden Prairie, MN]or a Polar Air convective air blanket [Augustine Medical]).The selection of blanket settings during both cooling andrewarming was at the discretion of the anesthesiologist ac-cording to their usual practice in this setting as needed toachieve target temperatures.

Perioperative Procedures

After the induction of anesthesia and placement of anesophageal temperature probe, the hypothermia-inducing de-vice (as assigned by randomization) was activated using set-

tings intended to achieve a target temperature of 33°C asrapidly as possible. As soon as hypothermia induction wasinitiated—but before target temperature was reached—eachpatient proceeded to craniotomy and aneurysm repair accord-ing to the neurosurgeon’s standard procedures and pace. Aclinical objective for the study was to evaluate whether pa-tients could achieve target hypothermia before the time theneurosurgeons applied the first temporary arterial or aneu-rysm clip when the surgeons were operating at their usualpace.

Near the end of surgery, after the final aneurysm clip was inplace or on the order of the neurosurgeon, active rewarmingwas begun by resetting the temperature settings on the endo-vascular device or the convective air blanket, according torandomization, as needed to rewarm the patient to normo-thermia as rapidly as possible. Active rewarming was to bediscontinued as soon as the esophageal temperature reached36.0°C. If normothermia was not achieved by the end of sur-gery, alternate warming measures could be used.

Throughout the surgical procedure, all patients underwentcontinuous temperature monitoring via the esophageal tem-perature probe. Patient temperatures and clock times wererecorded when active cooling was initiated and at selectedpoints during the surgical procedure.

Postoperative Management

After surgery, patients were managed according to usualinstitutional routine. In addition, all patients underwent phys-ical examinations (with a focus on the groin access site andclinical evidence of deep venous thrombosis or vascular in-jury) immediately after surgery and at 24 hours, 72 hours, and30 days thereafter. Venous Doppler ultrasound of the lowerextremities and femoral access site region was performedbefore surgery and at 72 hours after surgery in both groups toassess for any evidence of vascular injury or deep venousthrombosis. Laboratory assessments were performed at base-line, 24 hours after surgery, and 30 days after surgery. Neu-rological status was assessed at baseline (National Institutes ofHealth [NIH] Stroke Scale [2]) and at 30 days after surgery(Glasgow Outcome Scale, NIH Stroke Scale). Adverse eventswere evaluated through Day 30.

Statistical Analyses

The cooling rate was calculated as the degree decrease(Celsius) in temperature divided by the time (hours) measuredfrom the initiation of active cooling until the target tempera-ture was reached. If the target was not reached, time wasmeasured until active cooling was discontinued. Warmingrates were calculated similarly based on the time and temper-ature when active warming was initiated until a target tem-perature was reached or active warming was discontinued,whichever came first. The time to achieve target temperatureswas calculated from the time that active cooling was starteduntil the target temperature (33.0°C for cooling, 36.0°C forrewarming) was achieved. If this target was not reached, the

ENDOVASCULAR VERSUS SURFACE COOLING


time to target was estimated based on a linear projection of theobserved cooling or warming rate. Mean values were com-pared using Student’s t test; median values were comparedusing the Wilcoxon rank-sum test. The percentage of casesachieving mild hypothermia during surgery and achievingnormothermia by the end of surgery in each treatment armwere compared using ordinary �2 or Fisher’s exact tests. Anominal � level of 0.05 was used for these tests.

RESULTS

Patient Characteristics

From August 2000 through June 2002, 153 patients wereenrolled in the study at 15 clinical sites in the United States.Their mean age was 50 years (range, 28–70 yr), and their meanweight was 77 kg (range, 48–148 kg). Women constituted 71%of this study population.

Of the enrolled patients, 92 were randomized to undergoendovascular hypothermia and 61 were randomized to hypo-thermia induced with a surface convective air blanket. Onehundred forty-eight patients (97%) subsequently underwenthypothermia as randomized. In five cases (four in the endo-vascular group and one in the surface blanket group), theprotocol-specified hypothermia procedures could not be com-pleted, either because of technical issues with the respectivehypothermia devices or because of the discovery after enroll-ment of a condition that precluded use of the device. In onepatient randomized to endovascular hypothermia, femoralaccess for sheath placement could not be obtained. In twopatients randomized to endovascular hypothermia, hypother-mia could not be delivered because of a device malfunction. Inone patient randomized to surface blanket cooling, coolingwas interrupted and the blanket was removed to place afemoral sheath for an intraoperative angiogram, and the cool-ing blanket was not reapplied. Data from these five patientswere excluded from the analysis of cooling and warmingparameters.

Mean patient esophageal temperatures at baseline (begin-ning of active cooling) were similar in the endovascular and

surface blanket groups (35.2 and 35.5°C, respectively). Themean time from the initiation of hypothermia induction toplacement of the first temporary arterial or aneurysm clip alsowas the same for endovascular and surface blanket patients(152 and 153 min, respectively). For the endovascular patients,the mean inferior vena cava catheter dwell time (from catheterplacement to removal) was 274 minutes (range, 125–555 min).

Temperature Management Performance

Hypothermia

The hypothermia induction parameters for the two studygroups are summarized in Table 1. Compared with the surfaceconvective air blanket, the endovascular system providedmore rapid cooling with significantly shorter times required toreach target temperature. The target hypothermic temperatureof 33°C was achieved by the time the surgeon applied the firsttemporary arterial or aneurysm clip in only 20% of patients inthe surface blanket group, compared with 99% of patients inthe endovascular group (P � 0.001). Twenty-seven percent ofpatients in the surface blanket group were not able to attain atemperature of even 34°C by the time of first arterial or aneu-rysm clip placement.

Figure 2 depicts the mean patient temperature versus timeduring cooling (and warming) for those patients who hadcontinuous temperature readings recorded. The more rapidand extensive cooling, as well as the tight control of the targettemperature (�0.1°C) obtainable with the endovascular ap-proach compared with the surface blanket approach, werenoted.

Neither age nor sex seemed to affect the rate of coolingindependently (data not shown). Cooling rates varied in-versely with weight (and related parameters, such as bodymass index and body surface area). However, cooling rateswere consistently greater with endovascular cooling than sur-face cooling across all weight groups. Obese patients (definedas having a body mass index of �30 kg/m2) comprised 25% ofthe overall population in this study. Among these obese pa-tients (n � 22), the mean cooling rates were 3.56 (�1.50)°C/h

TABLE 1. Cooling parametersa

VariableEndovascular group

(n � 88)Surface blanket group

(n � 60)P value

Mean cooling rate (°C/h; � SD) 4.77 (� 2.07) 0.87 (� 0.54) �0.001

Mean time to 33°C (min; � SD) 34.8 (� 22.0) 204.4 (� 89.7) �0.001

By the time of first temporary arterial clip or aneurysm clip, the percentage ofpatients cooling to (%)

�33.0°C 99 20 �0.001�33.5°C 100 48 �0.001�34.0°C 100 73 �0.001

a SD, standard deviation.

STEINBERG ET AL.


for the endovascular group versus only 0.69 (� 0.33)°C/h inthe surface blanket group (P � 0.01). Among nonobese pa-tients (n � 66), mean cooling rates were 5.17 (� 2.09)°C/h forthe endovascular group and 0.97 (� 0.61)°C/h for the surfaceblanket group (P � 0.01). For obese patients, the target tem-perature of 33°C was achieved by first arterial or aneurysmclip placement in 96% of patients who underwent endovascu-lar cooling, but only in 5% of patients who underwent surfacecooling (P � 0.001).

Warming

The rewarming performance parameters in the two studygroups are summarized in Table 2. Compared with the surfaceconvective air blanket, the endovascular system providedmore rapid rewarming with significantly shorter times re-quired to reach target normothermia. Again, a significantlyhigher percentage of patients in the endovascular group re-warmed to 36°C by the end of surgery compared with patientsin the surface blanket group.

In addition, the percentage of patients still intubated onleaving the operating room was 28% for patients undergoing

temperature management with the surface blanket systemversus 14% for patients managed with the endovascular sys-tem (P � 0.035). Operative times were 239 � 77 minutes (mean� standard deviation) for the endovascular group and 249 �81 minutes for the surface blanket group.

Illustrative Case

A 58-year-old woman with a broad-necked, enlarging basilar arteryapex aneurysm underwent craniotomy with clipping of her aneurysm(Fig. 3, A and B). She was randomized to the endovascular group. Thetarget temperature of 33°C was attained by the time of the firsttemporary basilar artery trunk occlusion (Fig. 4). The aneurysm wasoccluded and the basilar bifurcation was reconstructed using aneu-rysm clips. There were no changes in the brainstem auditory evokedresponses or somatosensory evoked potentials, and microvascularDoppler examination demonstrated flow in the left P2 posterior cere-bral artery (PCA), as well as in the right P1 PCA. The patient wasrewarmed to 34.8°C (over 50 min) while an intraoperative angiogramwas obtained (Fig. 4). The angiogram disclosed an unanticipatedocclusion of the left P1 PCA by one of the aneurysm clips (indicatedby the arrow in Fig. 3C). Therefore, the patient was recooled to 33.7°C(over 10 min) for additional temporary clipping of the basilar arterytrunk and repositioning of the permanent clips. A repeat intraopera-tive angiogram demonstrated clipping of the aneurysm and filling ofboth the P1 and P2 PCAs (Fig. 3D).

Safety Parameters

No patient in the endovascular group developed any symp-tomatic or clinically significant catheter-related complications,such as deep venous thromboses. The screening postoperativeultrasound examination performed in all patients revealed anasymptomatic nonocclusive thrombus at some location in thelower extremity venous system in 0% of patients in the surfaceblanket group and in 2 (2.2%) of 92 patients in the endovas-cular group. In one of these patients, an initial postoperativeultrasound examination detected an asymptomatic nonocclu-sive femoral vein thrombus near the catheter insertion site, buta follow-up ultrasound study, performed a few days later,failed to confirm any lesion. In the second patient, an asymp-tomatic nonocclusive venous thrombus was observed in theperoneal—but not in the femoral—vein after several angio-

FIGURE 2. Graphs showing the mean body temperature versus time dur-ing the cooling and maintenance phases (left panel) and the warmingphase (right panel) for patients with continuous temperature recordingsin the endovascular group (n � 52; blue) and the surface blanket group(n � 18; orange). Bars, standard deviations.

TABLE 2. Rewarming parametersa

VariableEndovascular group

(n � 88)Surface blanket group

(n � 60)P value

Mean warming rate (°C/h; � SD) 1.88 (� 0.50) 0.69 (� 0.43) �0.001

Mean time to 36°C (min; � SD) 101.8 (� 29.1) 207.0 (� 103.0) �0.001

By the end of surgery, the percentage of patients warming to (%)�36.0°C 46 15 �0.001�35.5°C 69 27 �0.001�35.0°C 89 53 �0.001

a SD, standard deviation.



graphic procedures involving that lower extremity; the rela-tionship of this finding to the use of the device was unclear. Athird patient in the endovascular group was found on ultra-sound examination to have a small asymptomatic femoralarteriovenous fistula that required no intervention. No signif-icant bleeding or infectious complications at the femoral ac-cess site were reported.

Postoperative laboratory parameters were comparable inthe two study groups. Postoperative infection was reported in4.3% of endovascular patients and 4.9% of surface blanketpatients. This included one case of pneumonia, one case ofsepsis, one oral and vaginal fungal infection, and one ear andlymph node infection in the endovascular group, and one caseof pneumonia and two wound infections in the surface blan-ket group. All infections resolved with antibiotic treatment.Two additional patients in the surface blanket group devel-oped a postoperative fever of unknown cause that resolvedwith antibiotics, and one patient in the endovascular grouphad a postoperative fever that resolved spontaneously; thispatient had a negative workup. There were no significantdifferences between the two groups with regard to the post-operative Glasgow Outcome Scale or NIH Stroke Scale or inthe percentage of patients who showed a more than 4-pointworsening in their NIH Stroke Scale score (Table 3).

DISCUSSION

Mild brain hypothermia consistently has been shown to pro-vide excellent protection against both focal and global cerebralischemic injury in experimental animal models (3, 13–15, 18).Recently, two prospective randomized studies demonstrated thebenefit of mild hypothermia in improving neurological outcomein patients with cardiac arrest resulting from ventricular fibrilla-tion (1, 6). In the current study involving patients undergoingelective craniotomy for unruptured aneurysm repair, endovas-cular temperature management provided superior outcomes forall of the prespecified temperature management parameterscompared with surface cooling: cooling and warming rates, timeto achieve target temperatures, and percentage of patients

FIGURE 3. A and B, angiograms obtained before surgery showing abroad-necked, basilar artery apex aneurysm in a 58-year-old woman ran-domized to the endovascular group. C, intraoperative angiogram demon-strating occlusion of the left P1 PCA by one of the aneurysm clips(arrow). The patient was cooled again for additional temporary clipping ofthe basilar artery trunk and repositioning of the permanent clips. D,repeat intraoperative angiogram demonstrating clipping of the aneurysmand filling of both P1 and P2 PCAs (arrows).

FIGURE 4. Graph illustrating the temperature control during aneurysmsurgery for an illustrative patient (58-year-old woman; height, 168 cm;weight, 85 kg). The target temperature of 33°C was attained by the time ofthe first temporary basilar artery trunk occlusion (first arrow, “tempclip”). The aneurysm was occluded, and the basilar bifurcation was recon-structed using aneurysm clips. The patient was rewarmed to 34.8°C over50 minutes while an intraoperative angiogram was obtained (secondarrow, “angiogram”). The patient was cooled again for temporary clip-ping of the basilar artery trunk (third arrow, “temp clip”) and reposition-ing of the permanent clips, followed by rewarming.

TABLE 3. Neurological outcomesa

Endovasculargroup

Blanketgroup

Glasgow Outcome Scale (Day 30)GOS 5 85% 84%GOS 4 or 5 94% 96%

NIH Stroke Scale (Day 30)b

NIHSS 0 or 1 86% 84%

a GOS 5, good recovery, able to return to work or school; GOS 4, moderatedisability, able to live independently, but unable to return to work orschool.b According to Brott et al. (2).

STEINBERG ET AL.


achieving target temperatures during both hypothermia and ac-tive rewarming.

The higher cooling rates attainable with endovascular cool-ing permitted surgeons consistently to achieve desired hypo-thermia levels by the time of first arterial or aneurysm clipplacement. This was not often achieved for patients undergo-ing surface cooling. Of note, with endovascular cooling, rapidcooling was achieved even in obese patients, who can bedifficult to cool with surface cooling methods.

The ability to rapidly warm patients to normothermia by theend of surgery also is important for several reasons. A numberof deleterious effects have been documented in patients whoare hypothermic after surgery. These include a longer anes-thesia recovery time (11), an increase in postoperative bleed-ing (8, 17), an increase in surgical wound infection (10), anincrease in length of hospital stay (10, 17), an increase inmorbid cardiac events (4), and an overall increase in mortality(8). Failure to achieve normothermia by the end of surgeryalso may result in the inability to extubate the patient beforeleaving the operating room. For example, in one clinical studycomparing hypothermia with normothermia in neurosurgerypatients (5), the percentage of patients still intubated on leav-ing the operating room was 42% for patients assigned tointraoperative hypothermia versus 28% for patients main-tained at normothermia. In the present study, patients in theendovascular group rewarmed more quickly than patients inthe surface group. A 50% reduction in the number of patientsstill intubated on leaving the operating room was observed forpatients undergoing endovascular temperature managementcompared with those undergoing surface temperature man-agement. This likely reflects the significantly higher warmingrates and greater likelihood of reaching normothermia by theend of surgery seen in this study with endovascular warmingcompared with surface warming. Even higher rewarmingrates (approximately 2.7°C/h) have been achieved with anenhanced version of the same endovascular device (unpub-lished data).

The endovascular approach provided superior temperaturemanagement compared with surface convective air blankets,but it also required placement of a central venous catheter,which could increase the risk of catheter-related complica-tions. Therefore, thorough evaluation for any possiblecatheter-related complications was emphasized in this study,and an ultrasound evaluation of the lower extremities for allenrolled patients was performed to screen for even subclinicalvascular abnormalities. Although 2 of 92 patients in the endo-vascular group had ultrasound evidence of asymptomaticnonocclusive lower extremity vein thrombus and one patienthad a small asymptomatic arteriovenous fistula not requiringtreatment, extensive evaluation failed to reveal any clinicallysignificant catheter-related lesions.

The placement of a central venous endovascular device alsorequires additional operating time. However, this relative dis-advantage may be outweighed by the more rapid coolingtime, the ability to achieve and maintain target cooling tem-peratures, faster rewarming time, and an improved rate of

extubation in the operating room. Thus, this study suggeststhat intraoperative hypothermia can be induced, maintained,and reversed more efficiently via an endovascular approach,compared with a surface approach, without a significant in-crease in complications. Whether mild hypothermia improvesthe neurological outcome in patients undergoing aneurysmsurgery is not answered by this study, nor is the question ofwhether cooling to 33°C provides equivalent or additionalprotection compared with barbiturate-induced electroen-cephalographic burst suppression. However, if cooling is re-quired, the endovascular technique is preferred to the surfaceblanket method. Endovascular cooling may have clinicallybeneficial applications for inducing rapid and controlled mildhypothermia in patients undergoing cerebrovascular surgery,as well as in patients with acute stroke, head trauma, or acutemyocardial infarction.

DISCLOSURE

GKS, CSO, ESC, and SLG are members of the ScientificAdvisory Board for Innercool Therapies. TEB-S is a consultantto Innercool Therapies. RLA is an employee of InnercoolTherapies.

REFERENCES

1. Bernard SA, Gray TW, Buist MD, Jones BM, Silvester W, Gutteridge G,Smith K: Treatment of comatose survivors of out-of-hospital cardiac arrestwith induced hypothermia. N Engl J Med 346:557–563, 2002.

2. Brott T, Adams HP Jr, Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J,Holleran R, Eberle R, Hertzberg V, Rorick M, Moomaw CJ, Walker M:Measurements of acute cerebral infarction: A clinical examination scale.Stroke 20:864–870, 1989.

3. Busto R, Ginsberg MD: The influence of altered brain temperature in cere-bral ischemia, in Ginsberg M, Bougosslavsky J (eds): Cerebrovascular Disease:Pathophysiology, Diagnosis, and Management. Cambridge, Blackwell Science,1998, pp 287–307.

4. Frank SM, Fleisher LA, Breslow MJ, Higgins MS, Olson KF, Kelly S, BeattieC: Perioperative maintenance of normothermia reduces the incidence ofmorbid cardiac events: A randomized clinical trial. JAMA 277:1127–1134,1997.

5. Hindman BJ, Todd MM, Gelb AW, Loftus CM, Craen RA, Schubert A, MahlaME, Torner JC: Mild hypothermia as a protective therapy during intracra-nial aneurysm surgery: A randomized prospective pilot trial. Neurosurgery44:23–33, 1999.

6. Hyperthermia After Cardiac Arrest Study Group: Mild therapeutic hypo-thermia to improve the neurologic outcome after cardiac arrest. N EnglJ Med 346:549–556, 2002.

7. Inderbitzen B, Yon S, Lasheras J, Dobak J, Perl J, Steinberg GK: Safety andperformance of a novel intravascular catheter for induction and reversal ofhypothermia in a porcine model. Neurosurgery 50:364–370, 2002.

8. Insler SR, O’Connor MS, Leventhal MJ, Nelson DR, Starr NJ: Associationbetween postoperative hypothermia and adverse outcome after coronaryartery bypass surgery. Ann Thorac Surg 70:175–181, 2000.

9. Krieger DW, De Georgia MA, Abou-Chebl A, Andrefsky JC, Sila CA, KatzanIL, Mayberg MR, Furlan AJ: Cooling for acute ischemic brain damage (coolaid): An open pilot study of induced hypothermia in acute ischemic stroke.Stroke 32:1847–1854, 2001.

10. Kurz A, Sessler DI, Lenhardt R: Perioperative normothermia to reduce theincidence of surgical-wound infection and shorten hospitalization: Study ofWound Infection and Temperature Group. N Engl J Med 334:1209–1215,1996.



11. Lenhardt R, Marker E, Goll V, Tschernich H, Kurz A, Sessler DI, Narzt E,Lackner F: Mild intraoperative hypothermia prolongs postanesthetic recov-ery. Anesthesiology 87:1318–1323, 1997.

12. Mack WJ, Huang J, Winfree C, Kim G, Oppermann M, Dobak J, InderbitzenB, Yon S, Popilskis S, Lasheras J, Sciacca RR, Pinsky DJ, Connolly ES Jr:Ultrarapid, convection-enhanced intravascular hypothermia: A feasibilitystudy in nonhuman primate stroke. Stroke 34:1994–1999, 2003.

13. Maier CM, Steinberg GK (eds): Hypothermia and Cerebral Ischemia: Mecha-nisms and Clinical Application. Totowa, Humana Press, 2003.

14. Maier CM, Ahern K, Cheng ML, Lee JE, Yenari MA, Steinberg GK: Optimaldepth and duration of mild hypothermia in a focal model of transientcerebral ischemia: Effects on neurologic outcome, infarct size, apoptosis, andinflammation. Stroke 29:2171–2180, 1998.

15. Maier CM, Sun GH, Kunis D, Yenari MA, Steinberg GK: Delayed inductionand long-term effects of mild hypothermia in a focal model of transientcerebral ischemia: Neurological outcome and infarct size. J Neurosurg 94:90–96, 2001.

16. Ogilvy CS, Carter BS, Kaplan S, Rich C, Crowell RM: Temporary vesselocclusion for aneurysm surgery: Risk factors for stroke in patients protectedby induced hypothermia and hypertension and intravenous mannitol ad-ministration. J Neurosurg 84:785–791, 1996.

17. Schmied H, Kurz A, Sessler DI, Kozek S, Reiter A: Mild hypothermiaincreases blood loss and transfusion requirements during total hip arthro-plasty. Lancet 347:289–292, 1996.

18. Steinberg GK, Grant G, Yoon E: Deliberate hypothermia, in Andrews R (ed):Intraoperative Neuroprotection. Baltimore, Williams & Wilkins, 1995, pp 65–84.

AcknowledgmentsOther members of the Temperature Control during Aneurysm Surgery (TCAS)

Study Group include: Warren R. Selman, M.D. (University Hospitals of Cleveland/Case Western Reserve University School of Medicine); Stephen A. Cohen, M.D.(Beth Israel Deaconess Medical Center); John A. Wilson, M.D. (Wake ForestUniversity/Baptist Medical Center); Hunt Batjer, M.D., and Christopher Getch,M.D. (Northwestern Memorial Hospital/Northwestern University); Barbara Vande Wiele, M.D. (UCLA Medical Center); Joseph Zabramski, M.D. (St. Joseph’sMedical Center/Barrow Neurological Institute); Robert Singer, M.D. (University ofNebraska); and John Dobak, M.D., and Andrew Balo (Innercool Therapies). Thisstudy was supported in part by funding from Innercool Therapies. We thankHarriet Davis, Beth Hoyte, and Dave Schaal for assistance with the manuscript.

COMMENTS

This article describes a clearly improved method for induc-ing and reversing therapeutic hypothermia. The method

seems safe and effective and is much more rapid than surfacetechniques. The question is when the surgeon should be pre-pared for cooling using this technique. Should every patientwith potential vascular occlusion have the cooling catheterplaced in the vena cava, or is there a subset of patients under-going aneurysm repair or other vascular procedures in whomthis should be kept ready? Another question is whether thereare simpler methods that can offer the same degree of protec-tion as is obtained with moderate hypothermia. For instance,can barbiturate dosage adequate to induce burst suppressionpatterns on the electroencephalogram provide the same de-gree of protection? Knowledge of the answers to these com-parative questions will allow the most appropriate and safestcare for the patient to be delivered. This article is an excellent

and careful evaluation of a superior method for inducinghypothermia.

Charles J. Hodge, Jr.Syracuse, New York

Temporary clipping is often necessary during aneurysmsurgery. However, ischemic injury is always a threat, de-

pending on the location and duration of the clip placementand the presence or absence of collaterals. The pattern ofcollaterals is usually not firmly understood at the time ofsurgery for any given vessel; therefore, any temporary clipplacement longer than several minutes (the ischemic thresholdfor interruption of cerebral blood flow) carries the potential forischemic injury. Rates of radiographic evidence of ischemicinjury after temporary clipping range from 9.8 to 26% (1, 3, 4),and clinical stroke rates range from 5.25 to 17% (1, 3). Methodsto improve the safety of temporary clipping are needed.

The authors report a randomized trial comparing two meth-ods of induced hypothermia during craniotomy for unrup-tured aneurysms. They compared hypothermia induced by acentral venous endovascular device versus that induced bystandard surface cooling blankets. They examined rates ofcooling and rewarming and complication rates related to in-duced hypothermia. Cooling and rewarming rates were dra-matically shorter in the endovascular group, and the percent-age of patients reaching the target body temperature of 33°Cby the time of the first temporary clip placement was signifi-cantly higher in the endovascular group. Although minorlower-extremity complications occurred in three patientstreated with the endovascular device, infection rates for eachgroup were similar. Neurological outcomes also seemed to besimilar.

Induced hypothermia is gaining increasing attention as amethod to protect against cerebral ischemic injury in a varietyof clinical settings. A previous randomized trial of surfacecooling to induce hypothermia for cerebral protection duringaneurysm surgery was burdened by an inability to achieve thetarget temperature in some patients, particularly obese pa-tients, and a relatively long time to rewarm (2). In the presentstudy, the authors have shown that endovascular cooling istechnically superior to surface cooling. To determine whetherthis method is also superior in protecting against neurologicalinjury during aneurysm surgery will require a trial involvingseveral hundred patients. We look forward to a larger-scalestudy in the future to determine whether this neuroprotectivestrategy can decrease ischemic injury rates and improve clin-ical outcomes.

Mark R. HarriganL. Nelson HopkinsBuffalo, New York

1. Ferch R, Pasqualin A, Pinna G, Chioffi F, Bricolo A: Temporary arterialocclusion in the repair of ruptured intracranial aneurysms: An analysis of riskfactors for stroke. J Neurosurg 97:836–842, 2002.

STEINBERG ET AL.


2. Hindman BJ, Todd MM, Gelb AW, Loftus CM, Craen RA, Schubert A, MahlaME, Torner JC: Mild hypothermia as a protective therapy during intracranialaneurysm surgery: A randomized prospective pilot trial. Neurosurgery 44:23–33, 1999.

3. Ogilvy CS, Carter BS, Kaplan S, Rich C, Crowell RM: Temporary vesselocclusion for aneurysm surgery: Risk factors for stroke in patients protectedby induced hypothermia and hypertension and intravenous mannitol admin-istration. J Neurosurg 84:785–791, 1996.

4. Thome C, Vajkoczy P, Horn P, Bauhuf C, Hubner U, Schmiedek P: Contin-uous monitoring of regional cerebral blood flow during temporary arterialocclusion in aneurysm surgery. J Neurosurg 95:402–411, 2001.

The neuroprotective effective of hypothermia in experimen-tal cerebral ischemia is known. Although it has been

shown in a prospective randomized trial that mild hypother-mia is not of significant benefit, many neurosurgeons stillbelieve in its neuroprotective effect during aneurysm surgery.

In the study presented, the authors investigate endovascu-lar versus surface cooling for induction of hypothermia duringclipping of unruptured cerebral aneurysms in a prospectiverandomized trial. A total of 153 patients were enrolled in thestudy. The goal of the study was to achieve 33°C when the firstarterial or aneurysm clip was placed. The results of the studydemonstrate that endovascular cooling is superior to surfaceblanket cooling with respect to induction, maintenance, andreversal of hypothermia. Complications do not increase.

This is a very well-performed study that is of major clinicalimportance. Because the study was performed as a prospec-tive randomized multicenter trial, the data were achievedwithout selection bias and are statistically reliable. Importantquestions, e.g., cooling time, rewarming time, cooling of obesepatients, and complications, are considered. The authors pro-vide detailed information about the method. One might arguewhether hypothermia was really necessary in all patients. Theselection of patients who really need hypothermia duringsurgical clipping of the aneurysm might be the critical ques-tion. However, this was not the focus of the study. The mes-sage is clear: endovascular cooling is an effective procedurethat can be applied in difficult aneurysms without provokinga major risk for the patients.

Gabriele SchackertDresden, Germany

Endovascular cooling is surely an elegant way to control thepatient’s temperature quickly and accurately during sur-

gery. The advantages are clear compared with the blanketmethod. Regardless of the cost of the system described, the

potential value of clinical applications may be high in selectedcases. However, it has not been shown that hypothermia initself has any benefit in aneurysm surgery. Cooling, in fact,unintentionally prolongs surgery, especially in surgery forunruptured aneurysms, and there may be complications re-lated to that. Elective aneurysm surgery belongs in the handsof a skillful and experienced vascular neurosurgeon who canpotentially “guarantee” a quick and safe neurosurgical proce-dure, and then there might be no need for intraoperativecooling. Therefore, we ask the authors to publish the sites andsizes of the aneurysms treated once more.

Mika NiemeläKeisuke IshiiJuha HernesniemiHelsinki, Finland

Steinberg et al. have demonstrated the advantages of endo-vascular cooling, as compared with surface methods. This

was a multicenter, prospective, randomized study, which fol-lows previous studies using swine and nonhuman primates.We now know that rapid cooling and warming can beachieved by an endovascular system; we must move on tosolve the next questions. For example, are there disadvantagesin addition to the risk of deep vein thrombosis and the addi-tional operative time? Do rapid elevations and decreases inbody temperature damage the organs or muscles? In thisarticle, as well as in the authors’ previous study using swine,the investigators did not mention potassium and creatinephosphokinase data. And, more importantly, under what con-ditions is this advanced technique needed? In the presentstudy, there were no significant differences in neurologicaloutcomes between endovascular and surface treatmentgroups.

Disease-oriented studies will be the next step. Furthermore,as the authors mention, it remains unknown whether thistechnique is more advantageous than barbiturate-inducedcoma. Nevertheless, the result that an endovascular coolingsystem is efficiently and safely applied in human patients isimportant because this might lead to more selective braincooling, using smaller catheters, or even to selective heating oftumor feeders.

Jun B. TakahashiNobuo HashimotoKyoto, Japan



comparison of endovascular and surface cooling during unruptured cerebral aneurysm repair: the tcas...

Documents