Vous êtes sur la page 1sur 6


Rapid Communication
Hypothermic Nerve-sparing
Radical Prostatectomy: Rationale,
Feasibility, and Effect on Early Continence
David S. Finley, Kathryn Osann, Douglas Skarecky, and Thomas E. Ahlering
OBJECTIVES To report the first application of preemptive local hypothermia during robotic-assisted laparo-
scopic prostatectomy (hRLP) to attenuate inflammation. Surgical excision of the prostate during
radical prostatectomy causes inflammatory damage to the surrounding neuromuscular tissues that
AQ: 1 could affect urinary continence.
METHODS Of 50 consecutive patients undergoing nerve-sparing hRLP (case numbers 668-717; 3 were
excluded—2 underwent radiotherapy and 1 was withdrawn because of balloon failure), 47 were
prospectively compared with a standard RLP cohort (case numbers 1-667). Pelvic cooling was
achieved using cold irrigation and an endorectal cooling balloon cycled with 4°C saline. The
intracorporeal temperatures were measured. Continence was defined as 0 urinary pads. The
Kaplan-Meier analysis of the time to 0 pads and multivariate Cox proportional hazards regression
analysis was used to examine the group differences in continence after adjusting for the baseline
RESULTS The median temperature was 29.0°C (endorectal cooling balloon only, range 24.4°-35.9°C) and
25.5°C (endorectal cooling balloon plus irrigation, range 19.4°-34.0°C). The time to 0-pad status
was determined in 590 of 667 controls (88%). The 3-month hRLP 0-pad rate was 86.8% ⫾ 5.8%
and was 68.6% ⫾ 2.0% for the controls. The return to continence was faster for hRLP vs
controls: median 39 days (range 0-110) vs 59 days (range 1-720), respectively (P ⫽ .002, log-rank
test). A multivariate analysis adjusting for factors, including age, American Urological Associ-
ation symptom score, abbreviated International Index of Erectile Function-5, body mass index,
prostate weight, stage, nerve-sparing, and learning curve demonstrated a faster return to conti-
nence for the hRLP group relative to the control group (hazard ratio 1.66, 95% confidence
interval 1.11-2.49, P ⫽ .014).
CONCLUSIONS This study represents the initial application of local hypothermia to reduce the traumatic
inflammatory sequela of RLP. Hypothermia was easily induced and safe and resulted in a
statistically significant improvement in early postoperative continence. UROLOGY xx: xxx,
xxxx. © 2009 Elsevier Inc.

he incidence of urinary incontinence (ie, any to incontinence and in 64%-96% in combination with
pads) after radical prostatectomy (RP) in modern detrusor instability in 1 report.6 Other factors such as
series has ranged from 5.8% to 37.9%.1-3 Al- age, American Urological Association symptom score
though most men regain control, incontinence has a (AUAss), prostate weight, and erectile dysfunction have
major negative effect on their quality of life.4 RP-associ- been reported to play a role in the delay of the return
ated incontinence is thought to be related to a low continence.7-11
urethral closing pressure due to intrinsic sphincteric de- It is plausible that a contributing factor to transient
ficiency and, to a lesser extent, bladder dysfunction.5,6 incontinence is inflammation of the bladder and sphinc-
Urodynamically proven intrinsic sphincteric deficiency ter mechanism secondary to dissection of the bladder and
was reported in 35%-67% as the sole contributing factor prostate. Acute injury (due to dissection, traction, ther-
mal energy, etc.) and downstream sequelae of inflamma-
tion are linked to muscle and nervous tissue damage and
From the Department of Urology, University of California, Irvine, Medical Center and
Department of Medicine, University of California, Irvine, School of Medicine, Orange,
dysfunction due to cellular edema, acidosis, nerve con-
California duction blockade, free radical damage, and apoptosis.12-14
Reprint requests: Thomas E. Ahlering, M.D., Department of Urology, University of We hypothesized that preemptive local hypothermia to
California, Irvine, Medical Center, 33 City Boulevard West, City Tower, Suite 2100,
Orange, CA 92868-3298. E-mail: tahlerin@uci.edu the pelvis would attenuate the inflammation caused by
Submitted: August 18, 2008, accepted (with revisions): September 20, 2008 excision of the prostate and result in an earlier return to
© 2009 Elsevier Inc. 0090-4295/xx/$34.00 1
All Rights Reserved doi:10.1016/j.urology.2008.09.085
(T.A.). The standard pertinent clinical data, such as age, body
mass index (BMI), abbreviated version of International Index
of Erectile Function (IIEF-5) score, AUAss, urinary bother
score, prostate weight, prostate-specific antigen level, Gleason
score, clinical T stage, and use of nerve sparing, were entered
prospectively into an electronic database (Table 1). The insti- T1
tutional review board approved the study, and Health Insurance
Portability and Accountability Act compliance was observed.
The normothermic group was stratified into subgroups to
control for the learning curve and evolution of specific changes
in technique: cases 1-250, cases 251-450, cases 451-592, and
cases 593-667. In addition to being the most recent group of
patients, cases 593-667 mark the initiation of posterior recon-
struction as described by Rocco et al.10 In the hRLP group
O (cases 668-717), we did not change our technique (ie, the
L Rocco suture); the only change was the addition of hypother-
O mia.
Figure 1. Schematic depicting endorectal cooling balloon. Continence Assessment
Postoperative continence was assessed by addressed and stamped
landmark postcards (given to patients at their preoperative ap-
continence. To achieve locoregional hypothermia, we pointment). The patients returned the postcards as they achieved
implemented a novel method that represents the first the landmark (eg, landmark 1, 1 pad/d, landmark 2, pad free).
application of hypothermia to radical prostatectomy. We Continence was defined as the use of 0 pads.
report the feasibility data and our early continence re-
sults. Statistical Analysis
The data were analyzed by a nonclinical third party (K.O.) from
the Department of Medicine at the University of California,
MATERIAL AND METHODS Irvine, School of Medicine. The baseline characteristics be-
tween the 2 groups were analyzed with 2-group 2-tailed t tests.
Intraoperative Cooling Technique The primary clinical outcome was the interval to no urinary pad
Local hypothermia was achieved by devising an endorectal use. The Kaplan-Meier method was used to compare the inter-
F1 cooling balloon system (ECB) (Fig. 1). A 40 cm, 24F, 3-way val to no pads, and the log-rank statistic was used to test for
latex urethral catheter was placed inside a 5 ⫻ 2.5-in. elliptic significant differences. Cox proportional hazards regression analysis
latex balloon that distended at low pressure and conformed to was used to compare continence between the 2 groups after ad-
the rectal wall without excessive deformation. Anatomically, justing for baseline characteristics, including age, AUAss, IIEF-5
the ECB was designed to extend from the membranous urethra score, BMI, prostate weight, nerve sparing use, clinical stage, and
to the seminal vesicles. The lubricated ECB was inserted just learning curve, that could affect the return to continence. The
inside the anus and anchored by inflating the catheter balloon analysis software used was Statistical Analysis Systems (SAS In-
to 20 mL. The ECB was then distended and cycled continuously stitute, Cary, NC), with statistical significance considered P ⬍ .05.
with cold saline (4°C) by way of gravity, at 40 cm above the
patient. The ECB volume was approximately 200 mL and was
designed to be well below the known maximum isobaric dis- Complications
tension volume and diameter of approximately 315 mL and 6.2 The operative complications attributable to the technique, such
cm, respectively.15 Adjunctive 4°C sterile water intracorporeal as bleeding, hematuria, and gastrointestinal complaints, were
irrigation was used to augment cooling of the ECB. recorded. Patients were specifically queried about postoperative
A 9F esophageal probe (Smiths Medical ASD, Rockland, bowel symptoms (ie, proctalgia) using expanded prostate cancer
MA) was used to obtain intracorporeal temperature readings index composite questions 22-30. AQ: 2

directly along the anterior surface of the rectum/neurovascular

bundle. Intracorporeal temperature readings were captured dur-
ing 2 intervals: the first phase starting 10 minutes before the
prostatic vascular pedicle was transected and discontinued just Feasibility
before the completion of the urethral transection. The hypo- At no point was there a significant deviation (⬎1°C)
thermia was discontinued to spatially facilitate the apical dis- from the standard core body temperature in any of the
section and anastomosis. The second phase of cooling was patients. The intracorporeal temperatures were assessed
reinstituted after the anastomosis and continued until just be-
with the ECB alone and with cold intracorporeal irriga-
fore the patient was extubated.
A total of 50 consecutive men underwent hypothermic nerve-
tion (4°C). With ECB only, the median temperature was
sparing robotic-assisted laparoscopic prostatectomy (hRLP). Two 28.30°C (range 17.5°-35.4°C, standard deviation [SD]
patients were excluded because of previous radiotherapy that 3.54, 95% confidence interval [CI] 1.04). When adjunc- AQ: 3
had failed and 1 patient was withdrawn from the study because tive cold irrigation was used, the median temperature
of technical difficulties associated with the ECB (the ECB decreased to 25.10°C (range 18.0°-30.0°C, SD 2.93, CI
failed). All procedures were performed by a single surgeon 0.86; P ⬍ .0001). The median nadir temperature

2 UROLOGY xx (x), xxxx

Table 1. Baseline characteristics of hypothermia and control groups
Characteristic Control Group (Cases 1-665) hRLP (Cases 666-718) P Value (t Test)
Age (y) 61.3 ⫾ 7.4 59.9 ⫾ 6.9 .200
AUAss 8.7 ⫾ 7.2 7.7 ⫾ 6.2 .386
IIEF-5 19.0 ⫾ 7.5 22.7 ⫾ 2.9 .001
Preoperative PSA level (ng/mL) 6.7 ⫾ 6.2 6.0 ⫾ 6.4 .463
BMI (kg/m2) 26.9 ⫾ 3.4 26.5 ⫾ 2.8 .524
Prostate weight (g) 52.0 ⫾ 21.2 54.4 ⫾ 18.6 .463
Stage .236
I 386 (65) 26 (58)
II 187 (32) 17 (38)
III 17 (3) 2 (4)
Nerve-sparing technique .549
None 34 (6) 0 (0)
Unilateral 159 (27) 13 (28)
Bilateral 396 (67) 34 (72)
hRLP, hypothermia with robotic-assisted laparoscopic prostatectomy; AUAss, American Urological Association symptom score; IIEF-5,
abbreviated version of International Index of Erectile Function; PSA, prostate-specific antigen; BMI, body mass index.
Data presented as mean ⫾ SD or numbers of patients, with percentages in parentheses.

achieved was 21.0°C (range 15.0°-29.3°C, SD 3.29, CI p=0.002
0.97). Endorectal cooling was typically done for approx- 0.7

Pad-free (%)
imately 60 minutes during a standard case and 90 minutes 0.6
when lymph node dissection was performed. 0.5
Locoregional hypothermia added minimal operative 0.3
time. The estimated blood loss in the hypothermia group 0.2
was 86.5 mL (range 50-150) and was not significantly 0.1
different from that for the control group. Of the 50 0 50 100 150 200
patients, 1 complication occurred that was unrelated to Days
cooling and required intensive care unit admission for 24 A
hours of observation owing to an antiemetic extrapyra-
1 p=0.009
midal side effect. No rectal complaints occurred.
Outcomes—Urinary Continence 0.7
Pad-free (%)

Case #1-250
The interval to 0-pad status was determined in 47 of 47 0.6 Case #251-450
patients in the hRLP group (100%) and 590 of 667 0.5 Case #451-592
F2 controls (88%). Figure 2A demonstrates the Kaplan- 0.4 Rocco
0.3 hRLP
Meier analysis of the interval to 0 pads. Continence
returned significantly faster in the hypothermia group 0.2

(median 39 days) compared with the entire control group 0.1 C

0 O
(median 59 days, P ⫽ .002, log-rank test), representing a 0 50 100 150 200 L
33.9% improvement in the interval to continence. At 3 Days O
months, 86.8% ⫾ 5.8% of the hRLP group and 68.6% ⫾ R
2.0% of the control group were pad free.
On univariate analyses that included all patients, the Figure 2. Kaplan-Meier analysis of interval to 0 pads. (A)
median interval to 0 pads increased significantly with Normothermic (cases 1-667) vs hypothermia cohort (cases
668-717). (B) Cases 1-250, 251-450, and 451-594, Rocco
older age, greater AUAss, increasing prostate weight, and
technique vs hypothermia with robotic-assisted laparo-
T2 decreasing IIEF-5 score (P ⱕ .01 for each; Table 2). BMI, scopic prostatectomy.
clinical stage, and the use of nerve sparing were not
associated with the return to continence on the Kaplan-
Meier analyses. To address potential differences between 1.66 (95% confidence interval 1.11-2.49, P ⫽ .014),
the 2 groups in baseline characteristics, bladder neck- indicating a faster return to continence in the hRLP
sparing technique, and stage and to adjust for the inde- group relative to the control group.
pendent effects of these covariates on the interval to To assess the potential effect of the learning curve, we
continence, multivariate analysis using Cox proportional compared the interval to 0 pads for different subgroups of
T3 hazards regression was performed (Table 3). Age, AUAss, consecutive patients over time using Kaplan-Meier anal-
IIEF-5 score, and BMI were significant independent pre- ysis (Fig. 2B and Table 2) and Cox regression analysis.
dictors of the interval to continence. After adjusting for Although the interval to continence differed significantly
all potential covariates, the hazard ratio for hRLP was between the consecutive subgroups on univariate analysis

UROLOGY xx (x), xxxx 3

Table 2. Kaplan-Meier univariate analysis for interval to 0 a faster pad-free status relative to the Rocco cohort
pads (hazard ratio 1.7, 95% confidence interval 1.009-2.918,
Median Interval P Value P ⫽ .046).
Variable to 0 Pads (Log-Rank Test)
Age (y) ⬍ .001
ⱕ56 31 COMMENT
57-61 60 It is logical that the effects of surgically induced acute
61-68 58 injury to the surrounding nerves and muscles might delay
⬎68 77 urinary continence. It is well established that cooling can
AUAss .001
ⱕ3 48 dramatically reduce the consequences of acute trauma
4-7 58 (ie, traction, ischemia) and the inflammatory cascade.
8-12 49 Cooling profoundly reduces an array of damaging effects
⬎12 63 such as microcirculatory impairment due to endothelial
IIEF-5 ⬍ .001 damage, activation of the coagulation cascade, leukocyte
ⱕ17 89
18-22 60 infiltration, free radical production, cytokine formation,
22-24 40 acidosis, apoptosis, release of proteolytic enzymes, and
25 35 tissue destruction.16-22 Conceptually, preemptive hypo-
Prostate weight (g) .011 thermia prepares tissues for damage by lowering their
ⱕ39 47 metabolic rate and oxygen demands. It has been shown
39.1-47.6 63
47.7-59.3 49 that for each degree the temperature is decreased, a 5%
⬎59 60 reduction in oxygen consumption is achieved.16 As a
BMI (kg/m2) .233 result, less lactate formation occurs, protein synthesis and
ⱕ24.4 56 cell signaling are preserved, and, most importantly, the
24.5-2026.5 51 inflammatory response is blunted.17,18
26.6-2028.8 49
⬎28.8 59 A substantial body of clinical research has been
Nerve-sparing technique .508 amassed dating to the early 1940s demonstrating the
No 49 value of hypothermia for ischemic events such as anoxic
Yes 56 brain injury and cardiac arrest.23 In a randomized control
Clinical stage .727 trial of patients undergoing coronary revascularization,
I 55
II 60 138 patients were randomly assigned to either deep hy-
III 35 pothermia (⬍28°C) or normothermia.24 Of the 68 in the
Learning curve .009 normothermic control group, 7 (10.3%) incurred neuro-
Cases 1-250 49 logic deficits compared with 0 of the 70 hypothermic
Cases 251-450 64 patients (P ⫽ .006). A similar reduction was reported by
Cases 451-592 63
Cases 593-667 59 Cambria et al.25 among patients undergoing thoracic AQ: 4
Cases 668-717 39 aortic aneurysm, a 3% rate of spinal cord injury in 61
Rocco stitch .195 patients treated with hypothermia induced by continuous
No (cases 1-592) 56 epidural infusion of 4°C saline compared with 23% in 55
Yes (cases 593-717) 49 matched controls (P ⬍ .001). We hypothesized that local
hRLP .002
No (cases 1-667) 59 hypothermia might confer a similar therapeutic benefit to
Yes (cases 668-717) 39 the neuromuscular components for urinary continence.
This represents the first application of preemptive,
Abbreviations as in Table 1.
controlled local hypothermia to radical prostatectomy.
We have demonstrated the feasibility and safety. To
(P ⫽ .009, log-rank test), no consistent trend was seen in evaluate the improved outcomes in a pilot study com-
the median interval to 0 pads across the groups to suggest pared with existing outcomes, the challenge is the po-
a learning effect. Furthermore, on multivariate analysis, tential for bias resulting from inadequate numbers, base-
no evidence was found of improved continence with line characteristics, the presence of the learning curve,
experience when the learning curve was measured as a and technical changes. To maximize the statistical
continuous variable by consecutive patient number. The power, we analyzed the interval to no pads in the hRLP
hazard ratio for the learning curve after adjusting for group compared with our entire group and demonstrated
other factors was 0.9996 (95% confidence interval 0.999- a highly statistically significant benefit to cooling. To
1.000, P ⫽ .116). Finally, we evaluated the control cases account for the differences in baseline characteristics, we
593-667, representing our most recent cohort and the performed multivariate Cox proportional hazards regres-
introduction of the posterior stabilization stitch advo- sion analysis and demonstrated a faster return to conti-
cated by Rocco et al.10 Adjusting for differences in base- nence for the hRLP group (hazard ratio 1.66, P ⫽ .014).
line characteristics for both groups using multivariate Next, we analyzed for surgical experience (ie, learning
Cox regression analysis, the hypothermic group achieved curve), comparing subgroups for the entire cohort of 667

4 UROLOGY xx (x), xxxx

Table 3. Cox regression multivariate analysis of covariates
Variable Coefficient Hazard Ratio 95% Confidence Interval P Value
Age ⫺0.0253 0.9750 0.9614-0.9888 ⬍ .001
AUAss ⫺0.0148 0.9853 0.9711-0.9997 .046
IIEF-5 0.0216 1.0218 1.0057-1.0383 .008
BMI ⫺0.0373 0.9634 0.9380-0.9895 .006
Nerve-sparing technique ⫺0.1236 0.8837 0.7316-1.0675 .200
Stage (I vs II vs III) ⫺0.0587 0.9430 0.7770-1.1445 .553
Prostate weight ⫺0.0029 0.9917 0.9922-1.0021 .256
Learning curve ⫺0.0004 0.9996 0.9990-1.0001 .116
hRLP (no vs yes) 0.5081 1.6621 1.1096-2.4898 .014
Abbreviations as in Table 1.

patients (ie, cases 1-250, 251-450, and so forth). On Acknowledgment. To Ralph V. Clayman for his brilliant
univariate and multivariate analyses, we found no evi- insight, feedback, and support and to Douglas Skarecky for his
dence of improved continence with experience (Fig. 2B). beautiful artwork.
The hRLP group, however, had a faster return of conti-
nence on both univariate and multivariate analyses com-
1. Burkhard FC, Kessler TM, Fleischmann A, et al. Nerve sparing
pared with the other subgroups. Finally, we introduced
open radical retropubic prostatectomy— does it have an impact on
into our practice the posterior urethral plate reconstruc- urinary continence? J Urol. 2006;176:189-195.
tion (Rocco technique), which also comprised our most 2. Touijer K, Eastham JA, Secin FP, et al. Comprehensive prospective
recent subgroup (control cases 593-667). In this analysis, comparative analysis of outcomes between open and laparoscopic
we again used multivariate Cox proportional hazards radical prostatectomy conducted in 2003-2005. J Urol. 2008;179:
regression analysis to adjust for baseline characteristics
3. Penson DF, McLerran D, Feng Z, et al. 5-Year urinary and sexual
and again demonstrated a faster return to continence for outcomes after radical prostatectomy: results from the Prostate
the hLRP group compared with the Rocco cohort (hazard Cancer Outcomes Study. J Urol. 2005;173:1701-1705.
ratio 1.7, P ⫽ .046). 4. Litwin MS, Melmed GY, Nakazon T. Life after radical prostatec-
A weakness of the present study was that although we tomy: a longitudinal study. J Urol. 2001;166:587-592.
5. Hammerer P, Huland H. Urodynamic evaluation of changes in
thoroughly reviewed the basic and clinical published data
urinary control after radical retropubic prostatectomy. J Urol. 1997;
(which is without question extremely strong), there was 157:233-236.
still much trial and error. We initially attempted hypo- 6. Ficazzola MA, Nitti VW. The etiology of post prostatectomy in-
thermia much as one would with partial nephrectomy. continence and correlation of symptoms with urodynamic findings.
However, the thermodynamics of the blood flow (in and J Urol. 1998;160:1317-1320.
out) overwhelmed this concept. Thermodynamics be- 7. Rogers CG, Su LM, Link RE, et al. Age stratified functional
outcomes after laparoscopic radical prostatectomy. J Urol. 2006;
came a central issue. How to cool the region? How to 176:2448-2452.
cool without obstructing the operative field? How cold? 8. Milhoua PM, Koi PT, Lowe D, et al. Issue of prostate gland size,
How long? Et cetera! In essence, the present study rep- laparoscopic radical prostatectomy, and continence revisited. Urol-
resents an intent-to-treat study in that optimal cooling ogy. 2008;71:417-420.
was assuredly not achieved in patients early in our expe- 9. Nandipati KC, Raina R, Agarwal A, et al. Nerve-sparing surgery
significantly affects long-term continence after radical prostatec-
rience. A strength of the present study was that it repre- tomy. Urology. 2007;70:1127-1130.
sents level 1B evidence as defined by the Oxford Centre 10. Rocco F, Carmignani L, Acquati P, et al. Restoration of posterior
of Levels of Evidence (ie, individual cohort with ⬎80% aspect of rhabdosphincter shortens continence time after radical
follow-up). The continence data were obtained by vali- retropubic prostatectomy. J Urol. 2006;175:2201-2206.
11. Wei JT, Dunn RL, Marcovich R, et al. Prospective assessment of
dated self-administered questionnaires, limiting physician-
patient reported urinary continence after radical prostatectomy.
related bias. As discussed, the controls for the analysis were J Urol. 2000;164:744-748.
extensively evaluated for learning curve, technique, and 12. Ni Choileain N, Redmond HP. The immunological consequences
baseline characteristics. Whether the long-term continence of injury. Surgeon. 2006;4:23-31.
rates will be affected by more effective hypothermia awaits 13. Osborn L. Leukocyte adhesion to endothelium in inflammation.
Cell. 1990;62:3-6.
longer follow-up.
14. Weiss SJ. Tissue destruction by neutrophils. N Engl J Med. 1989;
15. Gladman MA, Dvorkin LS, Scott SM, et al. A novel technique to
CONCLUSIONS identify patients with megarectum. Dis Col Rectum. 2006;50:621-
The present study represents the first application of ac- 629.
tive targeted hypothermia to RLP against secondary in- 16. Yeh CH, Wang YC, Wu YC, et al. Ischemic preconditioning or
flammatory damage. hLRP resulted in a statistically sig- heat shock pretreatment ameliorates neuronal apoptosis following
hypothermic circulatory arrest. J Thorac Cardiovasc Surg. 2004;128:
nificant increase in the early postoperative continence
rates. Clearly, additional studies (at multiple centers) are 17. Sadgrove M, Pringle AK, Chad JE, et al. Intraischemic hypothermia
needed to evaluate the depth and duration of hypother- reduces free radical production and protects against ischaemic insults
mia to confirm our findings and optimize this technique. in cultured hippocampal slices. J Neuralchem. 2004;91:327-336.

UROLOGY xx (x), xxxx 5

18. Kawamura N, Schmelzer JD, Wang Y, et al. The therapeutic window cryotherapy attenuates microcirculatory impairment, regional in-
of hypothermic neural protection in experimental ischemic neuralpa- flammation, and muscle necrosis after closed soft tissue injury in
thy: protection in ischemic phase and potential deterioration in later rats. Am J Sports Med. 2007;35:93-102.
reperfusion phase. Exp Neurall. 2005;195:305-312. 23. Polderman KH. Induced hypothermia and fever control for preven-
19. Westermann S, Vollmar B, Thorlacius H, et al. Surface cooling tion and treatment of neurological injuries. Lancet. 2008;371:1955-
inhibits tumor necrosis factor-alpha-induced microvascular perfu- 1969.
sion failure, leukocyte adhesion, and apoptosis in the striated mus- 24. Mora CT, Henson MB, Weintraub WS, et al. The effect of temper-
cle. Surgery. 1999;126:881-889. ature management during cardiopulmonary bypass on neurologic and
20. Deal DN, Tipton J, Rosencrance E, et al. Ice reduces edema. J Bone neuropsychologic outcomes in patients undergoing coronary revascu-
Joint Surg. 2002;84:1573-1578. larization. J Thorac Cardiovasc Surg. 1996;112:514-522.
21. Laing DR, Dalley DR, Kirk JA. Ice therapy in soft tissue injury. N 25. Cambria RP, Davison JK, Zannetti S, et al. Clinical experience
Z Med J. 1973;78:155-158. with epidural cooling for spinal cord protection during thoracic and
22. Schaser KD, Disch AC, Stover JF, et al. Prolonged superficial local thoracoabdominal aneurysm repair. J Vasc Surg. 1997;25:234-241.

6 UROLOGY xx (x), xxxx