Neoplasm

The Effects of Hydrogen Peroxideon Brain and Brain TumorsAli H. Mesiwala, M.D., Lindi Farrell, M.S., Paul Santiago, M.D., Saadi Ghatan, M.D., andDaniel L. Silbergeld, M.D.Department of Neurological Surgery, University of Washington, Seattle, Washington

Mesiwala AH, Farrell L, Santiago P, Ghatan S, Silbergeld DL. Theeffects of hydrogen peroxide on brain and brain tumors. SurgNeurol 2003;59:398–407.

BACKGROUNDHydrogen peroxide (HP) is routinely used during neuro-surgical procedures to augment hemostasis after intra-cranial tissue resection. Elsewhere in the body, HP isused to kill resection margin tumor cells; in vitro studiessupport these clinical uses. The effects of HP on brain andbrain tumors have not been evaluated. In this study, thein vitro and in vivo effects of HP on both rat and humanbrain parenchyma and brain tumors were examined.

METHODSAntitumor activity of varied concentrations of HP (0–30%) on cultured primary and metastatic brain tumors (n� 13) was compared with the effects of various concen-trations of ethanol (0–50%). Studies were performed inrats to characterize HP-induced tissue changes that oc-curred when HP-soaked pledgets were placed on thearachnoid surface and along resection margins (n � 5).Additionally, the effect of HP on human brain along tumorresection cavities was investigated (n � 10).

RESULTSWhile HP demonstrated concentration-dependent tu-moricidal effects in vitro, similar to results achieved withethanol, HP caused significant injury to arachnoid andstroma with neuronal and glial injury to a depth of 1 mmin rats. Three percent HP-soaked pledgets placed in re-section cavities of excised brain tumors induced similarinjury in human brain.CONCLUSIONHP irreversibly damages mesothelial and neural tissue.Although HP appears to have tumoricidal effects in vitro,it should be used with caution in humans because of risksof collateral injury to surrounding normal brain. HP mayprove most beneficial for discrete lesions, such as pitu-itary tumors and metastases. © 2003 Elsevier Inc. Allrights reserved.

KEY WORDSBrain, human, hydrogen peroxide, neurosurgery, tumors.

Hydrogen peroxide (HP) is routinely used byneurosurgeons as an aid to hemostasis after

intracranial tissue resection [4,17] (e.g., after tumorresection, lobectomy, transsphenoidal pituitary tu-mor resection, and brain resection in the traumasetting). HP has also been used intraoperatively bysurgeons elsewhere in the body to kill tumor cells inthe resection margin (breast, liver, etc.), but thispractice is not widely accepted by neurosurgeons.Alternatively, ethanol has been used for extraopera-tive and intraoperative embolization and sclerosisof central nervous system tumors despite its knowntoxicity to neural structures.

Ethanol (ETOH) is commonly used by neurosur-geons after resection of pituitary tumors to kill re-sidual tumor [6,11,12,24]. Experience with extra-CNS tumors has demonstrated similar tumoricidalactivity for ETOH. Transcutaneous, intratumoral in-jection of ETOH has been used to treat hepatocel-lular carcinoma [20], benign thyroid nodules [29],and thymic cysts [7]. Preoperative intravascularETOH tumor embolization has been used for CNS[13] and extra-CNS tumors [8]. ETOH has also beenused for pituitary neuroadenolysis to relieve intrac-table cancer pain [19,22]. Although these tech-niques have been advocated for many years [6,11],there are no clinical or experimental studies sup-porting these practices. Furthermore, there may besignificant risk to CNS structures, such as the opticapparatus when this technique is used in the pres-ence of a recognized or unrecognized CSF leak [24]during pituitary lesion resection.

In vitro, HP has been shown to inhibit or kill giantcell tumors, breast cancer cells, and osteoblastswhile having little effect on normal fibroblasts inculture [16,18]. Despite this apparent selective ac-tivity against tumor cells, similar studies have notbeen performed for primary or metastatic braintumors. We evaluated the effects of HP on tumorcells in vitro and on rat and human brain in vivo.

Address reprint requests to: Dr Daniel L. Silbergeld, Department ofNeurological Surgery, 1959 N.E. Pacific Street, Box 356470, Seattle, WA98195-6470.

Received May 22, 2002; accepted November 12, 2002.

0090-3019/03/$–see front matter © 2003 Elsevier Inc. All rights reserved.doi:10.1016/S0090-3019(03)00029-6 360 Park Avenue South, New York, NY 10010–1710

Materials and MethodsESTABLISHMENT OF IN VITROHUMAN TUMOR CELL CULTURESAll procedures were approved by the University ofWashington Human Subjects Committee, and hu-man tissue was collected only after informed con-sent had been given.

Surgical specimens were obtained during routinesurgery for resection of primary, metastatic, andpituitary tumors in adult patients. The resectionand standard microneurosurgical methods werenot altered in any way for this investigation. Spec-imens were split, with one part reserved for culture(described below) and the remainder sent for rou-tine histopathological evaluation.

The techniques for culturing and maintainingbrain tumor cell lines have been previously re-ported [21]. Briefly, fresh surgical specimenswere transported from the operating room in ster-ile Dulbecco’s Modified Eagle Medium (DMEM);minced to a slurry with crossed scalpels; centri-fuged at 250� g at 4°C for 10 minutes; and resus-pended in calcium- and magnesium-free Hank’sBalanced Salt Solution (HBSS) containing atrypsin-EDTA 0.25% for 3 minutes at room tem-perature. The cell suspension was then washed inDMEM via centrifugation at 250� g at 4°C for 12minutes and resuspended in DMEM with 10%heat-inactivated fetal calf serum (FCS); 10,000�g/mL penicillin; 10,000 �g/mL streptomycin sul-fate; and 25 �g/mL amphotericin B. A viable cellcount, using the trypan blue exclusion (TBE)method, was performed and 10 ml of cell suspen-sion plated in 75 mL Falcon T-75 tissue cultureflasks at a density of 2.5 to 5 � 105 cells/ml. Cellswere incubated at 37°C in a humidified 5% CO2

atmosphere. Cells were stored in growth mediumwith 5% dimethylsulfomethoxide (DMSO) solu-tion in liquid nitrogen for later use. Table 1 sum-marizes the characteristics for these cell lines.

ESTABLISHMENT OF RAT TUMORCELL LINESRat tumor cells from the GH1 and GH3 cell lines(known to secrete growth hormone) [23] frozen ingrowth media with 5% DMSO were obtained fromthe American Type Culture Collection (Manassas,VA). These cells were thawed, washed, and grownin DMEM with 15% heat-inactivated FCS; 10,000�g/mL penicillin; 10,000 �g/mL streptomycin sul-fate; 25 �g/ml amphotericin B; and plated in T-75

culture flasks for incubation in a 5% CO2 humidifiedatmosphere at 37°C. Cells were stored in growthmedium with 5% DMSO in liquid nitrogen for lateruse. Table 1 summarizes the characteristics forthese cell lines.

EXPOSURE OF CULTURED CELLS TOETOH AND HPTumor cells from subconfluent monolayer cultures(exponential growth phase) were harvested bytrypsinization, resuspended in HBSS, and countedby TBE. Suspensions of 1.0 � 105 tumor cells wereprepared by serial dilution and pelleted by centrif-ugation. The cell pellets were exposed to 1cc ofeither ETOH or HP at varying concentrations for 5minutes. Total cell counts and viable cell countswere performed using TBE. Cell survival (100� livecells/total cell count) was calculated. Each datapoint represents the average of six experimentrepetitions.

Of note, for three of the glioma cell cultures(DLS143, NS1081, and C6), 24-hour post-HP expo-sure cell survival was determined. For these exper-iments, exposure to 3% HP was performed for 5minutes; then the cells were washed with HBSS. Thecells were then incubated in growth medium asdescribed above for 24 hours and then assayed forviability.

1 Specimen Characteristics for In Vitro Ethanol andHydrogen Peroxide Exposure Studies

GliomasWU11.1 Glioblastoma multiformeDLS143 Glioblastoma multiformeNS1081 Glioblastoma multiformeNS1087 Low grade oligodendrogliomaNS1105 Gemistocytic astrocytomaC6a Glioblastoma multiforme

Metastatic LesionsHTB27 Breast adenocarcinoma

(to brain)CRL6602 Mammary adenocarcinoma

(to brain)HTB81 Prostate carcinoma

(to brain)Pituitary Lesions

UW108 Pituitary adenoma-insulin-likegrowth factor 1

DLS152 Pituitary adenoma—growthhormone

GH1* Pituitary adenoma—growthhormone

GH3* Pituitary adenoma—growthhormone

aRat cell line.

399Hydrogen Peroxide in Brain Surg Neurol2003;59:398–407

IN VIVO HP EXPOSURE—RATAll animal procedures were approved by the Uni-versity of Washington Animal Care and Use Com-mittee. Adult Wistar rats were anesthetized withpentobarbital sodium (Abbott Laboratories, NorthChicago, IL) 50 mg/kg intraperitoneally (IP). Skullswere exposed by a longitudinal incision in the mid-line and a unilateral craniectomy was performedwith a high-speed drill (Dremel, Racine, WI). Thedura was opened with a scalpel, and a small corticalresection was performed. Either a sterile normalsaline (NS)-soaked, sterile 3% HP-soaked, or 30%HP-soaked cotton pledget was then placed over theentire cranial defect, including the cortical resec-tion cavity, to achieve hemostasis. After 5 minutes,the pledget was removed and the brain surfacegently irrigated with sterile NS. Animals were sac-rificed immediately thereafter with an overdose ofpentobarbital IP. Of note, for 3% HP, two additionalanimals were sacrificed 24 hours and 96 hours, re-spectively, after their HP exposure. Perfusion, brainremoval, fixation, sectioning, and mounting on glassslides were performed as previously described [28].Table 2 summarizes the lesion characteristics forthe in vivo study.

IN VIVO HP EXPOSURE—HUMANAll procedures were approved by the University ofWashington Human Subjects Committee, and hu-man tissue was collected only after informed con-sent had been given.

After gross total resection of primary and meta-static brain tumors, hemostasis was achieved usingstandard neurosurgical techniques, excluding theuse of HP. Biopsies were then taken of the resectionmargin and labeled “pre-HP.” These samples wereimmediately placed in paraformaldehyde, then sub-sequently processed, mounted, sectioned, andstained using standard histologic techniques. Afterverifying that hemostasis had been achieved, 3%

HP-soaked cotton balls were placed along the re-section margin and left in place for 5 minutes. Afterthe cotton balls were removed, additional biopsyspecimens were taken from the resection margin.These were labeled “post-HP,” placed in parafor-maldehyde, and processed as above. Tissue speci-mens were then examined by two blinded observ-ers for any evidence of damage or change.

ResultsEXPOSURE OF CULTURED TUMORCELLS TO ETOHTumor cell lines (n � 13, Figure 1) exposed to ETOHshowed decreasing surviving cell fraction with in-creasing ETOH concentration. One hundred per-cent cell kill was achieved at 50% ETOH after5-minute exposure for all tumors except one pitu-itary adenoma line (Figure 1C). As shown in Figure1, the LD50 for gliomas (n � 6, Figure 1A), metastatictumors (n � 3, Figure 1B), and pituitary adenomas(n � 4, Figure 1C) was 20% ETOH. Overall, pituitarytumors appeared to be the most resistant to ETOH(Figure 1C).

EXPOSURE OF CULTURED TUMORCELLS TO HPTumor cell lines exposed to HP also showed de-creasing survival with increasing hydrogen perox-ide concentration (n � 13, Figure 2). One hundredpercent cell kill was achieved at 30% hydrogen per-oxide after 5-minute exposure. The LD50 for gliomaswas 9% HP (Figure 2A), metastatic tumors 12% HP(Figure 2B), and pituitary adenomas between 20 to30% HP (Figure 2C). Of note, pituitary tumors werethe most sensitive cell type to 3% HP, with approx-imately 70% survival (Figure 2C).

For the three glioma cell lines that were allowedto grow for 24 hours after 3% HP exposure, no cells

2 Lesion Characteristics for In Vivo Hydrogen Peroxide Exposure Study

PATIENT HISTOLOGIC DIAGNOSIS LOCATION

1 Glioblastoma multiforme Left temporal lobe2A Metastatic breast carcinoma (lesion 1) Right parieto-occipital lobes2B Metastatic breast carcinoma (lesion 2) Right parieto-occipital lobes3 Anaplastic astrocytoma Left temporal lobe4 Low grade glioma Left partietal lobe5 Metastatic melanoma Left temporal lobe6 Anaplastic astrocytoma Right frontal lobe7 Oligodendroglioma Right frontal lobe8 Anaplastic oligo-astrocytoma Right frontal lobe9 Pituitary adenoma-growth hormone Pituitary gland

400 Surg Neurol Mesiwala et al2003;59:398–407

survived. Moreover, suspended cells treated with3% HP for 5 minutes were never able to adhere tothe culture well and underwent delayed death. Thesubconfluent cell cultures treated with 3% HP for 5minutes all died 24 hours postexposure and werenonadherent to the culture wells.

IN VIVO HP EXPOSURE—RATTreatment of intact cortex and resection marginwith 30% HP-soaked pledgets for 5 minutes resulted

in complete disintegration of brain parenchyma (n� 1, Figure 3 bottom panel), well beyond the resec-tion margin. Alternatively, treatment with 3% HP-soaked pledgets for 5 minutes resulted in mesothe-lial and parenchymal damage that spared vascularand endothelial structures, as seen by light micros-copy (n � 2, Figure 3 middle panel). Areas of ratbrain with intact arachnoid suffered mesothelialdamage characterized by disintegration or absence

1 Survival curves for tumorcells exposed to ETOH. A:

Glioma cells treated with in-creasing concentrations ofETOH exhibited decreased sur-vival. One hundred percent cellkill was seen at 50% ETOH, andthe LD50 was 20% ETOH. B: Met-astatic tumor cells exhibited asimilar survival curve to gliomacells. As noted, the LD50 was20% ETOH. C: Pituitary adenomacell lines appeared more resis-tant to ETOH, despite LD50 of20% ETOH and 100% cell kill at50% ETOH.

401Hydrogen Peroxide in Brain Surg Neurol2003;59:398–407

of cells in the arachnoid with preservation of thecollagenous connective tissue within this mem-brane. Damage along and up to 1 mm deep to theresection margin was highlighted by stromal vacu-olization and shrunken/degenerating neurons, as-trocytes, and microglia. Beyond this 1 mm rim, noevidence of damage was seen. Rats sacrificed 24hours (n � 1) and 96 hours (n � 1) post-3% HPexposure demonstrated identical results. In con-

trast, NS (n � 1, Figure 3 top panel) produced nosuch damage.

IN VIVO HP EXPOSURE—HUMANTreatment of the postsurgical resection marginwith 3% HP-soaked cotton balls yielded results inhuman tissue (n � 10, Figure 4) identical to thoseseen in rats with the exception of mesothelial dam-age, which was not assessed. Of the 10 pairs of

2 Survival curves for tumorcells exposed to HP. A:

Overall, glioma cells appearedto have intermediate sensitivityto HP compared with metastaticand pituitary adenoma cell lines(B, C). The LD50 for gliomas was9% HP. B: Metastatic tumor cellswere found to be the most resis-tant to HP with LD50 of 12%HP.Despite this, 100% cell kill wasachieved with 30% HP. C: Pitu-itary adenoma cell lines werethe most sensitive cell type toclinically available 3% HP with30% cell kill at this concentra-tion. Surprisingly, the LD50 forthese cells lay between 20 to30% HP, making them the mostresistant cells type overall.

402 Surg Neurol Mesiwala et al2003;59:398–407

specimens evaluated, three blinded examiners con-sistently identified the “pre” and “post” HP treatedsamples in eight specimen pairs. In the remainingtwo specimen pairs, tissue sampling artifacts pre-cluded proper identification.

DiscussionHydrogen peroxide has been used as an irrigationfluid or aid to hemostasis in neurosurgical proce-dures for many years [4,17]. The first published

report evaluating the in vitro effects of HP on neuraltissue was published in 1980 by Hansson and Vall-fors [5]. Primary cell cultures from newborn rathemispheres were exposed to different irrigationfluids used in neurosurgery including HP, Elliott’sartificial cerebrospinal fluid, normal saline, Ringer’ssolution, and culture medium 199 with Hank’s salt.Cell death occurred after incubation with 3% HP for5 minutes; all other solutions had no noticeableeffect on the viability of the cells. This study wasfollowed by in vivo studies on feline brain by the

3 In vivo response to HP—rat model. After a limited cortical resection of the right frontal brain using an 18-gaugeneedle, NS- or HP-soaked pledgets were placed in/over the resultant cavity for 5 minutes to aid in hemostasis. As

seen in the gross photographs, NS and 3% HP had little effect on the brain. Thirty percent HP resulted in a chemicalburn, dissolving not only the brain underlying the pledget but also any tissue that came in contact with HP runoff (notethe shrunken, pale appearance of the right cortex in the lower left panel). The corresponding medium power (�75)H&E-stained micrographs on right show the histologic effect of these solutions on brain in/along and deep to theresection margin. For NS, no tissue destruction was evident. 3% HP resulted in limited mesothelial and parenchymaldamage consisting of stromal disruption and shrunken cells. Severe parenchymal damage was caused by 30% HP, asevidenced by complete dissolution of parenchyma. Note that in the gross specimens, the left hemisphere was used ascontrol brain. In no case was damage seen in control tissue.

403Hydrogen Peroxide in Brain Surg Neurol2003;59:398–407

same group [25–27]. They found that surface irriga-tion with 3% HP for 5 minutes resulted in thrombo-sis of leptomeningeal vessels and extensive blood-brain barrier dysfunction through the cortex intothe white matter. Ringer’s solution and Elliott’s so-lution B were found to have little, if any effect, onthe meninges, cortex, white matter, or blood-brainbarrier; normal saline yielded slight blood-brainbarrier dysfunction throughout the exposed cortexand underlying white matter. Further evaluation ofmesothelial cell integrity (subdural and arachnoidsurfaces)occurred with scanning electron micros-copy. Five-minute exposure to 3% HP resulted incomplete disintegration of cells with preservationof the collagenous connective tissue. Normal salineproduced occasional mesothelial damage, whileRinger’s solution and Elliott’s solution B caused noeffects. It was concluded that HP was unsuitable forapplication to the brain surface in humans.

The results of our in vivo studies on rat brain andhuman tissue are similar to these prior studies; theexception, however, is that in our studies, HP-induced damage was limited to the treated arach-

noid surface and up to 1 mm of tissue beyond theresection cavity margin. In rats, HP produced com-plete cellular degeneration of the arachnoid thatwas subjected to HP while preserving the collage-nous connective tissue. In humans, brain paren-chyma suffered stromal vacuolization and degener-ation of neurons, astrocytes, and microglia up to 1mm beyond the resection margin when HP wasplaced in the resection cavity. No damage to vas-cular structures or endothelium was appreciated.These findings suggest that HP can be safely used inconfined spaces, such as brain tumor resection cav-ities and following pituitary microadenoma resec-tion where the likelihood of HP leaking into thesurrounding areas can be controlled and mini-mized. Clinically, HP has been safely used in thisexact manner for years; it is only when HP is al-lowed to diffuse outside the operative area into theventricles or used as an irrigant that complicationsmay occur.

There have been five reported cases of HP-relatedneurosurgical morbidity or mortality (Table 3)[1–3,14,15]. In all cases HP irrigation was used to

4 In vivo response to HP—human. Medium power (�75) H&E-stained micrographs of intraoperative biopsies ofresection margins taken from patients who had undergone surgery for tumor resection. The upper panels

correspond to the resection margin of a gross totally resected glioma, pre- and post-HP exposure. The bottom panelscorrespond to the resection margin after a transsphenoidal pituitary adenoma resection, pre- and post-HP exposure.As can be seen in the upper panels, treatment of the resection margin with 3% HP-soaked pledgets for 5 minutes yieldedstromal disruption and occasional shrunken cells. Similar findings are demonstrated in the lower panels where thesurface exposed to HP is seen as fragmented and containing shrunken cells.

404 Surg Neurol Mesiwala et al2003;59:398–407

augment hemostasis. Three cases [3,14,15] resultedin venous gas embolism characterized by hemody-namic collapse and multiple system failures; 2 pa-tients recovered [3,15], and 1 died [14]. In two othercases [1,2], HP irrigation resulted in severe poste-rior fossa pneumocephalus; 1 patient sufferedbrainstem herniation and expired [1], while theother underwent a decompressive suboccipitalcraniectomy and recovered [2]. Interestingly, inonly one case was the volume subjected to HPirrigation confined to a small space [2]; in this case,HP was instilled into a stereotactic biopsy site, re-sulting in the accumulation of gas, causing hernia-tion. In four of these five cases [1,3,14,15], the au-thors concluded that the practice of HP irrigationshould be avoided. In the other case [2], the authorconcluded that HP should only be used for “open”procedures and not for stereotactic operations. De-spite these rare cases of morbidity or mortalityassociated with the intraoperative use of HP, HPmay have a role as a tumoricidal agent in neurosur-gical procedures.

Our in vitro results indicate that HP, like ETOH,has tumoricidal effects in cultured cells. Thirty per-cent HP or 50% ETOH is adequate for 100% tumorcell kill at 5 minutes. As expected, higher concen-tration of HP or ETOH results in greater cytotoxic-ity. These results were obtained under ideal condi-tions: suspensions were a pure population of tumorcells not associated with vascular structures andcompletely covered with ETOH or HP. It appearsfrom this study that pituitary adenomas are moresensitive than either gliomas or metastatic tumorsto clinically available, sterile 3% HP. It is unclearfrom this study whether ETOH or HP is useful in thetreatment of other CNS tumors. Nonetheless, ourresults with a variety of tumor types are similar toprevious in vitro experiments with HP that demon-strated time- and concentration-dependent humanglioblastoma cell death [9]. In this study, investiga-tors found that the A172 human glioblastoma cellswere killed by HP exposure, and that this cell deathwas characterized by nucleosomal DNA fragmenta-tion and chromatin condensation. Additionally, p53protein was highly expressed and protein levels ofBak, p21WAF1/CIP1, and GADD45 were also en-hanced. Interestingly, cultured p53-deficient mouseastrocytes were more resistant to HP-induced apo-ptosis than wild-type astrocytes. These results sug-gest that repair of oxidatively damaged DNA maybe regulated by p53-induced p21WAF1/CIP1 andGADD45, and oxidative stress results in glial apo-ptosis mediated by p53-induced Bak. This theory isfurther supported by a recent study that demon-strated that HP triggers the up-regulation of Fas/3

Sum

mar

yof

Rep

orte

dC

omp

licat

ions

Seco

ndar

yto

Intr

aop

erat

ive

Hyd

roge

nP

erox

ide

Use

AU

TH

OR

SP

RO

CE

DU

RE

RE

ASO

NFO

RH

PU

SEC

OM

PLI

CA

TIO

NO

UT

CO

ME

Dol

an19

87[2

]St

ereo

tact

icb

iop

syof

cere

bel

lar

lesi

onIr

riga

tion

for

hem

osta

sis

Pos

teri

orfo

ssa

pne

umoc

eph

alus

,h

erni

atio

nV

enou

sga

sem

bol

ism

Surg

ical

pos

teri

orfo

ssa

dec

omp

ress

ion

wit

hre

cove

ry

Lub

nin

etal

1998

[15]

Cra

niot

omy

for

tum

orre

sect

ion

Irri

gati

onfo

rh

emos

tasi

sV

enou

sga

sem

bol

ism

Hem

odyn

amic

colla

pse

,m

ulti

syst

emor

gan

dis

ord

erw

ith

even

tual

reco

very

Dea

thLo

pez

etal

1999

[14]

Pos

teri

orsc

olio

sis

corr

ecti

onIr

riga

tion

for

hem

osta

sis

Ven

ous

gas

emb

olis

mD

eath

Dub

eyet

al20

00[3

]P

oste

rior

foss

asu

rger

yfo

rtu

mor

Irri

gati

onfo

rh

emos

tasi

sV

enou

sga

sem

bol

ism

Seve

reca

rdio

vasc

ular

colla

pse

follo

wed

by

reco

very

Ch

hab

raet

al20

00[1

]D

ebri

dem

ent

oflu

mb

ard

isc

spac

ein

fect

ion

Irri

gati

onfo

rh

emos

tasi

sP

oste

rior

foss

ap

neum

ocep

hal

us,

her

niat

ion

Dea

th

405Hydrogen Peroxide in Brain Surg Neurol2003;59:398–407

FasL expression in astrocytoma cell lines and aug-mented apoptosis [10]. These findings areconsistent with the apoptosis observed in cell cul-ture within 24 hours of 3% HP exposure. Whentaken together, the above may indicate that HP hasa role in neurosurgery following tumor resection;the selective, controlled treatment of resectionmargins with HP may result in the death of tumorcells within a 1 mm rim beyond the resection cavity,especially for pituitary adenomas.

ConclusionsThe intraoperative use of HP to augment hemosta-sis may not be as safe as previously believed. Theresults of our in vitro and in vivo studies indicatethat HP has clear antitumor effects in culture whilecausing mesothelial damage and stromal, neuronal,astrocytic, and glial degeneration to a depth of 1mm in in vivo specimens. Although HP appears tohave tumoricidal effects in vitro, it should be usedwith caution in the operative setting because of therisk of collateral injury. It may prove most impor-tant, and safest for small discrete lesion resections,such as those for pituitary tumors and metastaticbrain lesions.

This work was supported in part by KO8 NS01730 from theNational Institute of Neurologic Disorders and Stroke (DLS)and R29 CA69640 from the National Cancer Institute (DLS).

REFERENCES1. Chhabra R, Pathak A, Ray P. Fatal posterior fossa

pneumocephalus due to hydrogen peroxide irrigationof lumbar wound. Br J Neurosurg 2000;14:549–51.

2. Dolan EJ. Danger of use of hydrogen peroxide in ste-reotactic biopsies. Appl Neurophysiol 1987;50:237–8.

3. Dubey PK, Singh AK. Venous oxygen embolism due tohydrogen perozide irrigation during posterior fossasurgery. J Neurosurg Anesthesiol 2000;12:54–6.

4. Epstein JA. Hydrogen peroxide for hemostasis. Neu-rosurgery 1987;20:63.

5. Hansson E, Vallfors B. A study of irrigation fluids forneurosurgery on brain primary cell cultures. Experi-entia 1980;36:64–5.

6. Hardy J. Transsphenoidal Operations on the Pitu-itary. Randolph. MA: Codman and Shurtleff, 1982.

7. Hirano Y, Shimada T, Kinoshita Y, et al. Ethanol scle-rosis: one of the best treatments for thymic cyst invery elderly patients? Intern Med 1997;36:716–19.

8. Jungreis CA. Skull-base tumors: ETOH embolization ofthe cavernous carotid artery. Radiology 1991;181:741–3.

9. Kitamura Y, Ota T, Matsuoka Y, et al. Hydrogenperoxide-induced apoptosis mediated by p53 proteinin glial cells. Glia 1999;25:154–64.

10. Kwon D, Choi C, Lee J, Kim KO, Kim JD, Kim SJ, ChoiIH. Hydrogen peroxide triggers the expression of Fas/

FasL in astrocytoma cell line and augments apopto-sis. J Neuroimmunol 2001;113:1–9.

11. Landolt AM, Wilson CB. Tumors of the sella and para-sellar area in adults. In: Youmans JR, ed. NeurologicalSurgery, Vol 5, 2nd ed. Philadelphia: W. B. Saunders,1982:3107–62.

12. Laws ER Jr. Transsphenoidal approach to pituitarytumors. In: Schmidek HH, Sweet WH, eds. OperativeNeurosurgical Techniques. Indications, Methods andResults, Volume, 3rd ed. Philadelphia: W. B. Saunders,1995:283–92.

13. Lonser RR, Heiss JD, Oldfield EH. Tumor devascular-ization by intratumoral ETOH injection during sur-gery. Technical note. J Neurosurg 1998;88:923–4.

14. Lopez LM, Traves N, Napal M. Fatal gas embolismduring corrective scoliosis surgery using the poste-rior approach. Rev Esp Anestesiol Reanim 1999;46:267–70.

15. Lubnin AI, Sorokin VS. Gas embolism caused by erro-neous intraventricular administration of hydrogenperoxide solution. Anestexiol Reanimatol 1998;6:76–7.

16. Lucarotti ME, White H, Deas J, et al. Antiseptic toxic-ity to breast carcinoma in tissue culture: an adjuvantto conservation therapy? Ann R Coll Surg Engl 1990;72:388–92.

17. Mawk JP. Hydrogen peroxide for hemostasis. Neuro-surgery 1986;18:827.

18. Nicholson NC, Ramp WK, Kneisl JS, Kaysinger KK.Hydrogen peroxide inhibits giant cell tumor metabo-lism in vitro. Clin Orthop 1998;347:250–60.

19. Roberts MT, Henderson RS. Pituitary fossa injectionwith alcohol for widespread cancer pain. N Z Med J1981;93:1–3.

20. Sakamoto M, Hirohashi S. Natural history and prog-nosis of adenomatous hyperplasia and early hepato-cellular carcinoma: multi-institutional analysis of 53nodules followed up for more than six months and141 patients with single early hepatocellular carci-noma treated by surgical resection or percutaneousETOH injection. Jpn J Clin Oncol 1998;28:604–8.

21. Silbergeld DL, Ali-Osman F, Winn HR. Induction oftransformational changes in normal endothelial cellsby glioblastoma cells in culture. J Neurosurg 1991;75:604–12.

22. Takeda F, Uki J, Fujii T, Kitani Y, Fujita T. Pituitaryneuroadenolysis to relieve cancer pain—observationof spread of ETOH instilled into the sella turcica andsubsequent changes of the hypothalamopituitaryaxis at autopsy. Neurol Med Chir 1983;23:50–4.

23. Tashjian AH Jr., Yasumura Y, Levine L, et al. Estab-lishment of clonal strains of rat pituitary tumor cellsthat secrete growth hormone. Endocrinology 1968;82:342–52.

24. Tindall GT, Barrow DL. Prolactinomas. In Wilkins RH,Rengachary SS, eds. Neurosurgery Volume 1. NewYork: McGraw-Hill, 1985:852–8.

25. Vallfors B, Hansson HA, Svensson J. Experimentalexposure of the cat brain surface to neurosurgicalirrigation fluids and air: study of blood-brain barrierdamage. Neurosurgery 1980;7:53–6.

26. Vallfors B, Rosengren LE, Persson LI. Exposure of thecat brain surface to neurosurgical irrigation fluids,hydrogen peroxide and air. Quantitative assay of

406 Surg Neurol Mesiwala et al2003;59:398–407

blood-brain barrier dysfunction. Acta Neurochir(Wien) 1982;64:225–32.

27. Vallfors B, Hansson HA, Belghmaidi M. Mesothelialcell integrity of the subdural and arachnoid surfacesof the cat brain after exposure to neurosurgical irri-gation fluids and air: a scanning electron microscopicstudy. Neurosurgery 1983;12:35–9.

28. Wenzel HJ, Cole TB, Born DE, Schwartzkroin PA,Palmiter RD. Ultrastructural localization of zinctransporter-3 (ZnT-3) to synaptic vesicle membraneswithin mossy fiber boutons in the hippocampus ofmouse and monkey. Proc Natl Acad Sci USA 1997;94:12676–81.

29. Zingrillo M, Collura D, Ghiggi MR, Nirchio V, Tris-chitta V. Treatment of large cold benign thyroid nod-ules not eligible for surgery with percutaneous etha-nol injection. J Clin Endocrinol Metab 1998;83:3905–7.

COMMENTARYHydrogen peroxide is one of those agents that hasbecome ensconced in the lore of neurosurgery. Af-ter all, it must be of some benefit because of itseffervescence. I have convinced myself that thepresence of those bubbles is clear proof that apositive influence is being exerted on those bloodvessels that actually seem to be bleeding a bit lessafter peroxide has come into contact with them. Aspointed out in this paper, the peroxide solution

generates thrombosis and probably is the basis forits positive effect when the cavity in the brain isoozing and the frustration level is beginning tomount. Hydrogen peroxide oxidizes the tissues withwhich it comes in contact.

It has certainly been a staple of my armamentar-ium for these past 30 years. It has also been a stapleof my spouse’s armamentarium for at least thesame length of time. I must admit that I have neverused hydrogen peroxide for the purposes of influ-encing residual tumor. I have used peroxide to soakcotton balls in the tumor cavity or in a hematomacavity primarily for the purpose of controlling trou-blesome bleeding. My anecdotal experience is thatit is worthwhile. As pointed out, there is a risk whenthe solution is irrigated into the cavity and over thesurface of the brain; clearly, arachnoidal and corti-cal injury can be produced. For that reason, I haveused the 3% solution diluted by 50% with saline.Mesiwala and his colleagues have carefully evalu-ated the various effects of hydrogen peroxide andhave documented the pitfalls of indiscriminate use.

Harry O. Cole, M.D.Neurosurgeon

Chesterfield, Missouri

No one can see their reflection in running water. It is only instill water that we can see.

—Taoist Proverb

407Hydrogen Peroxide in Brain Surg Neurol2003;59:398–407