Embed Size (px)
Citation preview
1996;56:4509-4515. Cancer Res Michael Höckel, Karlheinz Schlenger, Billur Aral, et al. in Advanced Cancer of the Uterine CervixAssociation between Tumor Hypoxia and Malignant Progression
Updated version
http://cancerres.aacrjournals.org/content/56/19/4509
Access the most recent version of this article at:
E-mail alerts related to this article or journal.Sign up to receive free email-alerts
Subscriptions
Reprints and
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Permissions
To request permission to re-use all or part of this article, contact the AACR Publications
on June 3, 2014. © 1996 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from on June 3, 2014. © 1996 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from
(CANCER RHSl-AKCH 56. 45(»-45l5. Octnbcr I. 1996]
Association between Tumor Hypoxia and Malignant Progression in AdvancedCancer of the Uterine Cervix1
Michael Höckel,2Karlheinz Schienger, Billur Aral, Margarete Milze, Uwe Schäffer, and Peter Vaupel
Department iif Ohsieirics anil üyneciiltiuy /M. H., K. S.. B. A.. U. S.] unti lile Institutes ¡ifPiilholiifty ¡M.M./ und Ph\\inli>K\ ami Paihuiihysii>li>Ky¡P-V.]. University <ifMainzMetileni Schnei. 55/W Mitin:. Genniin\
ABSTRACT
Experimental tumors contain a significant fraction of microregions thatare chronically or transiently hvpoxic. Experimental evidence showingthat hypoxia (und suhsequent reoxygenation) muy huve a profound impacton malignant progression and on responsiveness to therapy is growing.I lie clinical relevance of tumor oxygénationin Immun solid malignancies
is under ¡mesticatimi.\\ e huve developed and validuted a clinically applicable method for
measurement of tumor oxygénation in locully udvunced cancer of theuterine cervix using a computerized polarographic electrode system. Applying this procedure in patients with cervical cancers S3 cm in diameter,who gave informed consent. \ve have been studying the clinicul relevanceof tumor oxygénationprospective!)' since 1989.
As of June 1995, 103 patients with advanced cancers of the uterinecervix [Federation Internationale des Gynaecologistes et Obstetristes(FICO) stages Ih. bulky di = 13). Ha and Uh (H = 51), IIIu and IHh(n = 34). and IVa und I\ h (n = 5)| hud entered the study. Fifty "i of the
patients had carcinomas with median pO, readings <10 mm Hg. referredto as hypoxic tumors.
Tumor oxygénation was found to be independent of various patientdemographics and also of pretreutment tumor churucteristics, such asclinical tumor stage and size, histológica! type, and differentiation. However, histopatbological examination of the surgical specimens followingradical tumor resection in 47 patients showed that lm\-p<>, tumors ex
hibited larger tumor extensions and more frequent (occult) parumetrialspread, as well as lymph-vusculur space involvement, compared to well-oxygenated tumors of similar clinicul stuge und size. Forty-two putients
completing primary radiatimi therapy and 47 patients who underwentradical surgery were analyzed for treatment outcome after u medianobservation period of 28 months (range. 3-76 months). Patients withhypoxic tumors had significantly worse disease-free und overall survival
probabilities compared to patients with nonhypoxic tumors. Cox regression analysis identified tumor oxygénationand FICO stage as the mostimportant independent prognostic factors. The poorer outcome of thepatients with hypoxic tumors was mainly due to locoregionul failures withand without distunt metustuses. irrespective of whether surgery or radiation was applied as primary treatment.
Tumor oxygénation as measured with a standardized polarographicmethod proved to be a powerful new pretherapeutic prognostic parameterproviding important information on malignant progression in terms ofextracervical tumor spread and rudioresistunce in advanced cervicalcancers.
INTRODUCTION
In normal tissue, oxygénation is a spatially and temporallyheterogenous homeostatic flow equilibrium determined by micro-regional oxygen delivery and oxygen consumption. In fast-growing
experimental tumors, a Multhusian principle characteri/ed by amore rapid increase of oxygen-consuming tumor cell populationsthan expansion of the oxygen-supplying functional microvascula-
Rcccivcd 4/5/96; accepted X/I6M6.The costs ot publication of this article were defrayed in part by the payment of page
charges. This article must therefore he hereby marked aihvriisement in accordance wilh18 U.S.C. Section 1734 solely to indicate this laci.
1Supported by Deulsehe Krehshille Grani M4(V9]/Va I.
To whom requests tor reprints should be addressed, at Universitätsfrauenklinik.Langcnhcckstrassc I. D-55IOI Main/. Cicrmany.
ture is evident ( I ). As a consequence, tissue hypoxia results fromeither insufficient O-, diffusion (chronic hypoxia) or insufficient
perfusion (acute hypoxia. transient hypoxia. or ischemie hypoxia).Moreover, in vitro and experimental tumor studies showed thathypoxia (and subsequent reoxygenation) may have a dramaticimpact on malignant progression in terms of tumor spread andresistance to therapy (2-13). Despite these striking observations in
tumor models, the clinical relevance of tumor oxygénation is stillunclear. After Thomlinson and Gray (14) presented evidence forhypoxic microregions within human tumors from a detailed his-
topathological study of lung cancers. Urbach (15) was the first tomeasure low oxygen tensions in malignant skin tumors directlywith pO2 polarography. Gatenby et al. (16) reported in 1988 theresults from polarographic oxygen tension measurements in lymphnode métastasesof head and neck cancers showing a significantrelationship between low mean intratumoral pO-, values and failure
to respond to fractionated radiotherapy thus supporting older studies, such as that of Kolstad (17). with cancer of the uterine cervix.These investigations, however, did not receive widespread attention because of the low number of patients involved, methodological limitations of pO, determinations, and both short and/or incomplete patient follow-up.
In 1989. a computerized histography system was introduced, allowing quick and reliable polarographic tissue pO, readings in theclinical setting (18-21 ). Applying this device in patients with locally
advanced cancer of the uterine cervix, we initiated a controlled prospective trial to evaluate the clinical relevance of tumor oxygénation.The measurements were performed according to the principles ofsystematic random sampling (22, 23) to take into account intratumorulheterogeneity. After having presented preliminary data from the firstanalysis in 1992 demonstrating the powerful predictive value of tumoroxygénation on recurrence-free and overall survival for patients
treated with radiation (24. 25). we now report the results from a largerpatient cohort and longer follow-up, including a subgroup of patients
who underwent primary surgery. We show that the adverse outcome of patients with hypoxic cervical tumors is independent of thestandard treatment and apparently represents a tumor biologicalphenomenon.
MATERIALS AND METHODS
Study Protocol. All patients with cervical cancers of at least 3 cm in largestdiameter as estimated by clinical investigation who had been admitted to theDepartment of Obstetrics and Gynecology at the University of Mainz MedicalSchool for treatment were eligible for the open prospective trial initiated inJune 1989. The study was approved by a medical ethics committee. Allpatients needed to give informed consent to be enrolled in the study. Clinicalstaging according to FIGO1 and si/ing were performed by experienced
gynecological oncologists.Histopathology. Histológica! type and grade according to the modified
Broders' system were obtained from routine H&E-stained sections of defined
tumor biopsies. The specimens from radical surgery were processed in giantsections for histológica! examination to verify tumor extension, depth of
1The abbreviation used is: FIGO. Federation Internationale des (¡ynaecologisles et
Ohslclrisles.
4509
on June 3, 2014. © 1996 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from
IIYPOXIA IS O-:RVICAI CANCI-.K
cervical invasion, and bladder or rectal involvement. Paramctriul tumor infiltration was investigated in the axial sections of the lateral parts of the surgicalspecimens. The presence of tumor cell clusters within an endothelial-linedspace was regarded as lymph-vascular space involvement. This feature was
classified as pronounced if it could he detected in nearly all of the high-power
fields at the invading tumor front. The demonstration of a tumor cell emboluswithin a blood vessel identified by the tunica media was interpreted asevidence of vascular invasion. Four sections from each removed lymph nodewere scrutinized for tumor métastases.
Treatments. Primary treatments with curative intent were performed uniformly during the study period by either surgery or radiation. Treatmentdecisions for the individual patients were made at a tumor conference withparticipation of the radiation oncologists and gynecological oncologists basedon the clinical judgment of tumor rescctability and on the age and nutritionaland performance status of the patients. The results of the tumor oxygénationmeasurements were not contributed and did not influence any decision ontreatment.
Surgical treatment consisted of radical hysterectomy or primary pelvicexenteration with pelvic lymph node dissection. Definitive radiation wasadministered as combined teletherapy and brachytherapy. External beamirradiation was applied with 10-MV photons produced by a linear accelerator at the Department of Radiology. For the brachytherapy. a high-dose-rate Ir 192 afterloading machine at the Department of Obstetrics and Gyn-ecology was used. The patients usually received 45 Gy whole-pelvisexternal-beam radiation with the standard four-field box technique at2-Gy/day fractions five times per week, followed by three to four endo-
cavitary brachytherapy insertions with a Hentschke applicator delivering atotal dose of 24-28 Gy at a dose rate of 0.5-1 Gy/min to point A at weekly
intervals. Lateral tumor spread was treated with additional parametrialteletherapy boosts of up to 15 Gy.
In contrast to the primary treatment, adjunctive therapies varied duringthe course of this study. From 1989 to 1992 a series of patients received"induction chemotherapy" with cisplatin. vincristine. and bleomycin fol
lowed by definitive radiation. During that time, surgically treated patientswith parametrial and nodal involvement underwent adjuvant chemotherapywith carboplatin and ifosfamide. From 1993 on. chemotherapy was givenconcomitantly with radiation using cisplatin or carboplatin instead of thesequential mode. Adjuvant treatment for surgical risk factors was changedto external-beam radiation of the pelvis with 50 Gy with the four-field boxtechnique at 2-Gy/day fractions.
The following treatments for recurrent disease have been applied: percutaneous pelvic radiation up to 60 Gy for postsurgical locoregional recurrences,pelvic exenteration for central pelvic disease, and the CORT procedure forside-wall disease in cases with previous pelvic radiation (26).
Tumor Oxygénation Measurements. Tumor oxygénationwas measuredpretherapeutically with the Eppendorf histography system (Eppendorf. Hamburg. Germany) adhering to the standard procedure as developed and validatedearlier (21. 24. 25). pOs readings were performed in the conscious patientalong linear tracks, first in the s.c. fat of the mons pubis followed by cervicalmeasurements at the 12 and 6 o'clock sites of the macroscopically vital tumor
tissue. Prior to 1993. pO2 measurements in the normal tissue were not carriedout. Within the tumor tissue, up to 35 pO, measurements were taken in eachtumor track (70 readings in total) starting at a tissue depth of 5 mm. Themeasuring points were placed 0.7 mm apart from each other, resulting in anoverall measurement track length of approximately 2.5 cm.
After the pO, measurements, needle core biopsies of approximately 2 mmin diameter and 20 mm in length were taken from those tumor areas where thepOs determination had been made and processed for histology. Both the pO,readings and the taking of the needle core biopsies were done without anesthesia in all patients. Intravaginal temperature, blood pressure, heart rate,hemoglobin concentration, hematocrit. and arterial oxyhemoglobin saturationwere monitored concomitantly with the pO, determinations. The pretherapeu-tic pO, measurements were usually performed 1-5 days prior to the oncolog
ical treatment. For validation of the standard procedure, a series of pO2measurements has been performed in the same patients (¡i)at two differenttimes, (ft) with two different histographs. and (c) by two different investigators.In another series, in addition to the measurements at the standard 12 and 6o'clock locations, up to eight further measurements at different sites were done
in the same tumor.
Follow-Up. Regular follow-up investigations were performed either in the
Department of Obstetrics and Gynecology at the University of Main/ MedicalSchool or in other clinical institutions at 3-month intervals during the first 2
years following treatment and twice yearly thereafter. The patients wereinterviewed and examined clinically. In case of symptoms or suspicious orpathological clinical findings, further diagnostic procedures were ordered,including interventional biopsies for definitive diagnosis. Further informationregarding the posttreatment disease course of the patients was obtained bytelephone interviews or correspondence with the patients, their relatives, ortheir primary care physicians.
Statistical Analysis. Statistical analyses were performed with the SASSottware. Version 6.04. For comparison, the determination of correlationcoefficients. Fisher's exact test, and the Mann-Whitney-Wilcoxon test (U test)
were applied. Survival and recurrence-free survival probabilities were calculated with the Kaplan-Meier method. Differences between survival probabilities were analyzed with the log-rank test. Variables influencing survival anddisease-free survival probabilities were evaluated with the univariate and
multivariate Cox proportional hazards model. A P value ^0.05 was consideredto indicate statistical significance.
RESULTS
From June 1989 until June 1995. 134 patients with advanced(i.e.. >3 cm in clinical diameter) cancers of the uterine cervix were
admitted to the Department of Obstetrics and Gynecology. Duringthat period, 23 eligible patients could not be accrued to the controlled prospective study because either the principal investigatoror the pO, histograph was not available. Three patients could notbe involved due to missing informed consent. In the remaining 108patients, pO-, measurements have been performed. The measure
ments were tolerated by all patients; 13 patients complained aboutminor discomfort. No complications occurred as a consequence ofthe pO2 determinations. Bleeding caused by the hypodermic needleprobe insertion was minimal, and bleeding from the needle corebiopsies did not exceed 20 ml in the majority of cases. Twopatients had to be treated with temporary intravaginal packings. Nopatient required blood transfusions for any blood loss due to thestudy protocol.
On the basis of the histológica! investigation of the core biopsies,intratumorul measurements were made in 103 of the 108 patients.Eight patients had more than two intrutumorul measurement tracks. 83patients had two, and 12 patients had only one.
To validate the procedure, we determined interobserver. intratu-
Tahlc I MclììiHÃŒDÃŒiì^ìi'itìf\'ÃititÃitii>nof ìtitnitiiiiitinil¡tO m̂t't¡\urt'iiu'nt\ in i
cancer nf ihr uterine cenïx
Intcrohservcrvariation"Standard
procedure,twoinvestigators,samehistographStandard
procedure,twoinvestigators,twohistographsStandard
procedure.measurementattwolimes
(At. 1-5days)tnlralumnral
variation'12o'clock v.v. 12o'clock6
o'clock i'.v.6o'clock12o'clock vs. 6o'clockStandard
v.v.extendedIntertumoralvariation'No.
oftumors1699302091g103DeviationMean2.31.82.73.55.04.50.98.9(mm
Hg)Range0-9.90.3-8.80.1-9.30-170-160-200.1-3.4O.l^tO.6
" The reference values for calculatin«!the deviations were the median p()2 from the
pooled measurements of the individual tumors.'The reference values were the median pO-, lor each measuring track.
' The reference value was the median pO-, of all tumors.
4510
on June 3, 2014. © 1996 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from
HYPOXIA IN CERVICAL CANCHK
Table 2 Preireatinent iunior characteristics and tumor oxygénation
No. of patients
VariablesClinical
tumorstageClinical
tumorsi/.eHistológica!
typeDifferentiationFIGO
Ib(bulky)FIGOIla.hFIGOIlla,hFIGOIVa.bmaximum
diameter ^5cmmaximumdiameter >5cmSquamouscellcarcinomaAdcnocarcinoma.
adenosquamouscarcinomaG1/2G
3Hypoxie
tumors725l'I125274663220Nonhypoxictumors62615425264293714InfluenceP(FIGO
I+PPPon
oxygénation=
1.000III.V.Ill=
1.000=
0.416=
0.296+
IV)
moral, and intertumoral variations. The results are compiled in Table1. If the standard two-track procedure was performed, the interob
server variation of the median pO, with respect to histograph, operator, and time of measurement was of the magnitude of only 2-3 mm
Hg. Intratumoral variation between measuring positions located just afew millimeters apart from each other was of the same order as thevariation between measuring positions, which spanned some centimeters. However, although intratumoral heterogeneity in microregionalpO, values was pronounced, intertumoral heterogeneity was significantly higher if compared with the U test, thus allowing the use of themedian pO, as determined with the standard procedure for characterization of individual cervical cancers.
For the 103 patients in the trial, the mean oxygénation(i.e.. themean of the medians of the pooled histograms) was 12.5 mm Hg.Fifty-two patients had tumors with median pCh <10 mm Hg. designated "hypoxic tumors," whereas the remaining 51 patients had tumors with median pO-, values >10, termed "nonhypoxic." Median
pO3 values correlated significantly with the relative frequencies of thetwo lowest pOi classes of the pooled histograms (median pO,, relativefrequency of measurements up to 2.5 mm Hg: r = —¿�0.7361,P = 0.0001: median pO2, relative frequency of measurements up to 5mm Hg: /•= -0.6071. P = 0.0001).
Tumor oxygénationwas independent of patient demographics suchas age. menopausal status, parity, smoking, and hemoglobin levels.Likewise, as shown in Table 2. tumor oxygénationwas independent ofthe pretreatment tumor characteristics, such as clinical tumor stageaccording to the FIGO, clinical tumor size, and histological type anddifferentiation (grading).
In 47 of the 103 patients, the tumors had been resected by radical
hysterectomy (/? = 43) or exenteration (n = 4) along with pelvic
lymph node dissection. In this subgroup, again, no correlation between tumor oxygénationand the clinical tumor data (FIGO stage andsize) or histological type and differentiation was found. However, thehistopathological investigation of the surgical specimens identifiedsignificantly larger maximum tumor extensions, higher pT categoriesdue to occult parametrial spread, and more frequent lymph-vascular
space involvement in the hypoxic tumors as compared to their nonhypoxic counterparts (Table 3).
Thirteen patients in whom tumor oxygénationmeasurements hadbeen performed were not treated for cure by standard therapy orrefused or abridged treatment. Table 4 presents a breakdown of thepretreatment tumor characteristics, treatment modalities, and tumoroxygénation status of the remaining 90 patients treated for cureaccording to accepted standards. No significant differences weredetected between the hypoxic and nonhypoxic tumors in the wholegroup, the radiation subgroup, or the surgery subgroup with respect toFIGO stage, clinical tumor size, and adjuvant treatment. One patientwas lost to follow-up. After a median follow-up time of 28 months(range, 3-76 months) 35 of the 89 patients progressed or relapsed.
Five of the relapsing patients received another treatment for cure.Thirty patients died of cervical cancer, and the death of one patientwas due to intercurrent disease (Table 5).
The overall and disease-free survival probabilities calculated withthe Kaplan-Meier method were significantly lower for the patients
with hypoxic cervical cancers compared to those with nonhypoxictumors (Fig. 1). The results of the regression analysis with the Coxproportional hazards model are given in Table 6. Tumor oxygénation,expressed as median pO,. was the strongest independent predictor of
Table 3 Hìstopalhaìoxiuillunwr chanicterixlic\ and Junior oxygénationof 47 patients treated with primary surgery
As shown in Table 4. no difference in clinical stages and si/es with respect to tumor oxygénationwas found for this subgroup of patients.
No. ofpatientsVariablePathological
tumor stage(pT)(Occult)
parametrialtumorextensionMaximum
tumorextensionLymph-vascular
spaceinvolvementVascular
invasionNodal
statusIb
(bulky)2a2h3a.
bYesNo^4
cm>4cmNegative/moderatePronouncedNegativePositiveNegativePositiveUnknownHypoxic
tumors(n =25)222012141114111414118170Nonhvpoxic
tumorsdi =22)7681913175175IS411101Influence onoxygénationP
=<pT,hi'P
=<PTlb/2aP
=P
=P
=P
=0.0627».
pT.,„>0.026>'•'•
PT2b»)0.0360.03620.06980.231
4511
on June 3, 2014. © 1996 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from
MVI'OMA IN CKRVK'AI. CANCliK
Table 4 Treatment modalities. FICO stages, and tumor o.iyxenaliun
TreatmentsubgroupPrimary
surgeryAdjuvantCT"Adjuvant
RTAdjuvantCT +RTFIGO
Ib(bulky)FIGOIla.hFIGO
IllaMaximumdiameter s5cmMaximumdiameter >5cmPrimary
RTInductionCTConcomitantCTFIGOIbFIGO
Ila.bFIGO1MbMaximum
diameter ^5cmMaximumdiameter >5 cmTotal471215312323331442II)S118231626Hypoxie
tumors256917Ih21872352o914716Nonhypoxictumors2266251611371»53199910
" CT, chemotherapy; RT. radiation therapy.
overall and disease-free survival in the study population of patients
with advanced cancer of the uterine cervix. Analyzing both subgroupsof patients treated with primary surgery and primary radiation revealed significant disadvantages in the outcome for the patients withhypoxic tumors irrespective of the mode of primary treatment (Figs.2 and 3). As demonstrated in Table 5, the majority of recurrencesappeared in the pelvis with or without a distant component, indicatingthat extracervical pelvic tumor spread and radioresistance were themost important reasons for treatment failure.
DISCUSSION
This prospective study shows that the pretherapeutic determination of the tumor oxygénation status provides important information concerning the malignant progression in terms of extracervicalspread and radioresistance of locally advanced cancer of the uterine cervix. Although knowledge about oxygénation in human tumors in situ may be obtained indirectly through various noninva-sive, highly sophisticated methods, such as magnetic resonancespectroscopy, single photon emission computed tomography, orpositron emission tomography to detect labeled hypoxia markers,we applied the invasive polarographic electrode assay since italone allows the direct measurement of tumor tissue pO2 valuesand has therefore been regarded as the "gold standard" [for a
review, see Stone et al. (27)].The pO2 measuring protocol applied in this trial allows for the valid
determination of tumor oxygénation within a few minutes in the
conscious patient without significant discomfort. Extending the number of measuring tracks from two (in the standard protocol) to up to10 did not significantly influence the results, but it prolonged theprocedure and is therefore not warranted. Similar experiences havebeen reported for head and neck lymph node métastasesand soft-
tissue sarcomas by others (28, 29). We restricted our investigation tocervical cancers of at least 3 cm (as estimated clinically) to enablevalid intratumoral measurements under clinical conditions. The potential to accurately determine tumor oxygénationin smaller tumorswith the Eppendorf histograph is currently being investigated in ourinstitution.
We used the median pO2 of all investigated cervical cancers,which was 10 mm Hg, as the cutoff level to discriminate hypoxicfrom nonhypoxic tumors. Interestingly, Vaupel el til. (30), in ananimal tumor model, have identified significantly different metabolic energy status in tumors with median pO2 <10 mm Hgcompared to those with median pO2 >10 mm Hg. Singer et al. (31)
demonstrated with breast cancer cell lines that diminished mitochondria! energy generation was related to malignant progression.Whether different metabolic states are also relevant for humancervical cancers with different oxygénation profiles remains to beestablished. In a preliminary investigation, associations betweenenergy status measured with a bioluminescence technique in random tumor samples and incidence of métastases were recentlyfound (32).
What determines the oxygénation status in cervical cancer? Itcan generally be assumed that the pO-, distribution in solid neo
plasms is influenced by host as well as tumor cell factors. In ourstudy cohort, we could not find any relation between tumor oxygénationand patient characteristics such as age, menopausa! status,smoking, hemoglobin concentration, and parameters of systemicblood circulation, which may affect oxygénationin normal cervicalstroma (21).
This study with more than KM)patients also confirmed our earlierfindings that tumor oxygénationin locally advanced cervical cancer isindependent from clinical stage, clinical tumor size, and histologicaltype and grading. However, contrary to the pretherapeutic clinicalfindings, the histopathological evaluation of the complete surgicalspecimens from 47 patients who underwent radical surgery revealedsignificantly larger maximum tumor extensions and more frequent(occult) parametrial infiltration, as well as pronounced lymph-vascu
lar space involvement at the tumor front, in hypoxic cervical cancersas compared with the nonhypoxic tumors. Clearly, tumor oxygénationmeasurements identified more aggressive tumor phenotypes, whichwere underestimated by the clinical evaluation with respect to size and
Table 5 Patterns of treatment failure and tumor oxygénati
DiseasestatusNED"RecurrencePelvisDistantPelvis
+distantTreatment
forrelapseExenterationCORTRTDeathFrom
cervicalcancerFromintercurrent diseaseHvpoxic
tumors(n =25)12135441TIO10Surgery
(n = 47)Nonhypoxic
tumors(n =22)1575111144Primary
radiation (;i =42)Hypoxic
lumors Nonhypoxic tumors(n = 23) (n =19)12
15II
49310i
iII
5II41
' NED, no evidence of disease; CORT. combined operative and radiotherapeutie treatment: RT, radiation therapy.
4512
on June 3, 2014. © 1996 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from
HYPOXIA IN CERVICAL CANCHR
A1.0-
n0.6,
pO2> 10 mmHg
n = 41
0.4
0.2-
0.0-
B
0.8,
0.61
pO,< 10mmHg
n = 48
Log-rank p = 0.0039
10 20 30 40 50Time [months]
60 70 80
10.4 H
0.2Hb
o.o-
•¿�i.4!.
pO,> 10 mmHg
n = 41
pO3 < lOmmHg
n = 48
Log-rank p = 0.0085
10 20 30 40 50Time [months]
60 70 80
Fig. I. Overall (A) and disease-free (ß)survival probabilities calculated with theKaplan-Meier method for 89 patients treated with curative intent stratified for tumor
oxygénation.
with low-pO2 and well-oxygenated tumors both for the radiation and
surgery treatment group, and therefore the comparison appears not tobe biased.
Why do hypoxic cervical cancers have a poorer outcome? Since themajority of the patients who were not disease-free relapsed in thepelvis with or without simultaneous distant métastases,radioresis-
tance and extracervical pelvic tumor spread must be considered asprimary causes of failure. Several mechanisms established from invitro studies and animal tumor models may aid to explain these
clinical findings.(a) The radiosensitizing effect of molecular oxygen for photon
irradiation should be diminished in low-pO2 tumors, which maycontain significant levels of radiobiologically hypoxic fractions (33-
35). However, direct measures of tumor oxygénationmay not berepresentative of the radiobiological hypoxic fraction, if a substantialproportion of cells within a tumor are nonclonogenic (36). Likewise,the impact of pretreatment tumor hypoxia for fractionated radiotherapy with claimed reoxygenation is unknown at present (37).
(b) Tumor cells, like normal cells, when starved in a hypoxicmicroenvironment can respond with the expression of a variety of
OverallsurvivalprobabilityA1.0-0.8-0.4-
0.2-~?I
Surgery*-*
Tt...
!_.»-•I«.
pO,>10mmHg«* n =22!;
ii
ipOj
< lOmmHg
n =25Log-rank
p = 0.01070
10 20 30 40 50 60 708Time[months]
stage. We are presently investigating the relation between histopatho-
logical size, vascularity, and oxygénationin cervical cancers. It appears that a Malthusian principle of neoplastic cell populations outgrowing their nutritive blood supply with increasing tumor size, whichhas been observed in experimental tumors (1), may also be relevant inhuman solid neoplasms.4
Five-year overall survival and disease-free survival probabilities
calculated for 89 patients who underwent standard primary treatmentfor cure were significantly lower for hypoxic tumors than for nonhy-
poxic tumors of similar clinical stages and sizes. Multivariate Coxregression analysis revealed tumor oxygénationas the most powerfulpretreatment prognosticator in our study cohort. Most interestingly,the disadvantage in outcome for low-pO2 tumors was independent of
the mode of primary treatment (radiation or radical surgery). Unlikethe uniform primary treatment, adjuvant treatments changed duringthe trial, and recommendations were not always followed by thepatients or their health care providers. As a result, there was nosignificant difference between the adjuvant treatments of the patients
4 M. Höckel.K. Schlenger. B. Aral, and P. Vaupcl. manuscript in preparation.
B
rä 0.6 H
0.4-
0.2-
Surgery
pO,> 10 mmHg
n = 22
10mmHg
O.O-1
n = 25
Log-rank p = 0.0804
10 20 30 40 50
Time [months]60 70 80
Fig. 2. Overall (A} and disease-free (ß)survival probabilities calculated with theKaplan-Meier method for 47 patients treated with primary surgery stratified for tumor
oxygénation.
4513
on June 3, 2014. © 1996 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from
HVPOXIA IN CERVICAL CANCER
Table 6 Cox regression analysis of the pretreattnent tumor characteristics of 89 patients with locally advanced cen'ical cancer treated with curative intent
VariableUnivariateTumor
oxygénation,hypoxic v.v.nonhvpoxic
FIGO stage. Ih. Ila, b vs. Illa,bMultivariateTumor
oxygénation,hypoxic vs.nonhvpoxic
FIGO stage. Ib. Ha. b 1-5.Ilia, bOverall
survivalOdds
ratio (95<7rCD"3.058(1.383-6.757)
1.945(0.936-4.042)P
value0.0059
0.07480.0039
0.0495Disease-free
survivalOdds
ratio (95%CD2.551
(1.201 -5.236)
1.504 (0.748 -3.025)P
value0.0108
0.25250.0084
0.0692"CI. confidence interval.
genes coding for oxygen-regulated proteins (38, 39). Several potentialmechanisms have been suggested for how some of these oxygen-
regulated proteins, including p53 and VEGF, influence tumor aggressiveness in terms of malignant progression and decreased responsiveness to therapy (2-10).
(f) There is evidence that hypoxia enhances genetic diversity inaddition to exerting a selective pressure (11-13). This may lead to the
0.8-
°6 -
0.4-
0.2
0.0-
Radiation
pO; > 10 mmHg
n = 19
10mmHgn = 23
Log-rank p = 0.0638
10 20 30 40 50 60 70 80Time [months]
Radiation
pO,>10mmHg
19
donai evolution of metastatic and intrinsically resistant tumor cellvariants according to Darwinian principles.
In a recent paper. Brizel et al. (40) showed in high-grade soft tissue
sarcomas a relationship between pretherapeutic tumor hypoxia and thedevelopment of distant métastasesas first site of failure followingtreatment with irradiation/hyperthermia and surgery. Although locore-
gional failures were dominant in our investigation of patients withadvanced cervical cancer, it is interesting to note that in the surgicallytreated subgroup of patients. 8 of 13 recurrences of the hypoxictumors had a distant component, as compared to only 2 of 7 recurrences in the nonhypoxic tumors.
In both cervical cancer and soft tissue sarcoma the hypoxic tumorphenotype is strongly associated with advanced malignant progression.This tumor biology phenomenon must be taken into consideration whenevaluating the effect of hypoxia on non-surgical anticancer treatments.
REFERENCES
1.
30 40 50Time [months]
Fig. 3. Overall {A) and disease-free (ß)survival probabilities calculated with the 14.Kaplan-Meier method for 42 patients treated with primary radiation stratified for tumor
oxygénation.
4514
Vaupel. P.. Kallinowski. F.. and Okunieff. P. Blood flow, oxygen and nutrient supply.and metabolic microenvironment of human tumors: A review. Cancer Res.. 49:6449-6465. 1989.Graeber. T. G.. Peterson. J. F.. and Tsai. M. Hypoxia induces the accumulation of p53protein, but the activation of a G,-phase by low oxygen conditions is independent ofp53 slalus. Mol. Cell Biol.. 14: 6264-6277. 1994.Giaccia. A. J. Hypoxic stress proteins: survival of the fittest. Semin. Radial. Oncol..f,: 46-58. 1996.diaeher. T. G.. Osmaninan. C.. Jacks. T.. Housman. D. li.. Koch. C. K.. Lowe. S. W..and Giaccia. A. J. Hvpoxia-mcdiated selection of cells with diminished apoptoticpotential in solid tumors. Nature (Lond.). 379: 88-91. 199ft.Plate. K. H.. Breier. G.. Weich. H. A., and Risau. W. Vascular endolhelial growthfactor is a potential tumor angiogenesis factor in human «liornas in rim. Nature(Lond.). 35V: 845-848. 1992.
Shweiki. D.. Itin. A.. Neufeld. G.. Gilaygoren. H., and Keshet. E. Patterns ofexpression of vascular endolhelial growth factor (VEGF) and VEGF receptors in micesuggesl a role in hormonally regulated angiogcnesis. J. Clin. Invest.. 91: 2235-2243.
1993.Brown. L. F.. Berse. B.. Jackman. R. W.. Tognazzi. K.. Manseau. E. J.. Dvorak. H. F..and Scnger. D. R. Increased expression of vascular permeability factor (VEGF) andils receptor in kidney and bladder carcinomas. Am. J. Pathol.. 143: 1255-1262. 1993.
Hlalky. L.. Tsionou. C.. Hahnfeldl. P.. and Colcinan. N. Mammary libroblasts mayinfluence breast tumor angiogenesis via hypoxia-induced vascular endothelial growthfactor up-regulation and protein expression. Cancer Res.. 54: 6083-6086. 1994.Toi. M.. Hoshina. S.. Takayanagi. T.. and Tominaga. T. Association of vascularendothelial growth factor expression with tumor angiogenesis and with early relapsein primary breast cancer. .Ipn. J. Cancer Res.. <V5:1045-1049. 1994.Takahashi. Y.. Kitadai. Y.. Bucana. C. D.. Cleary. K. R.. and Ellis. L. M. Expressionof vascular endothelial growth factor and its receptor. KDR. correlates with vascu-larily. metastasis, and proliferation of human colon cancer. Cancer Res.. 55: 3964-
3968. 1995.Young. S. D.. Marshall. R. S.. and Hill. R. P. Hypoxia induces DNA ovcrreplicationand enhances metastatic potential of murine tumor cells. Proc. Nail. Acad. Sci. USA.85: 9533-9537. 1988.Rice. G. C.. Hoy. C.. and Schimke. R. T. Transient In poxia enhances the frequencyof dihydrofolate reducÃasegene amplification in Chinese hamster ovary cells. Proc.Nati. Acad. Sci. USA. 83: 5978-5982. 1986.Russo, C. A., Weber. T. K.. Volpe. C. M.. Stoler. D. L.. Pelrelli. N. J.. Rodrigucv-
Bigas. M.. Burhans. W. C.. and Anderson. G. R. An anoxia inducible endonucleaseand enhanced DNA breakage as contributors to genomic instability in cancer. CancerRes.. 55; 1122-1128. 1995.Thomlinson. R. H.. and Gray. L. H. The histologie structure of some human lungcancers and the possible implications for radiotherapy. Br. J. Cancer. 9: 537-549.
1955.
on June 3, 2014. © 1996 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from
HYPOXIA IN ŒRVÕCAL CANO.R
15. Urhach, l;. Pathophysiology of malignancy: I. Tissue oxygénationof benign andmalignant tumors of the skin. Proc. Soe. Exp. Biol. Mcd.. V2: 644-649, 1956.
16. Galenhy, R. A.. Kessler, H. B., Rosenhlum, J. S.. Coiu, L. R., Moldofsky, P. J.. Hartz.W. H.. and Broder.G. J. Oxygen distributionin squamouscell carcinomamétastasesand its relationship to outcome ot"radiation therapy. Int. J. Radial. Oncol. Biol. Phys..
14: 831-838, 1988.17. Kolstad, P. Intercapillary distance, oxygen tension, and local recurrence in cervix
cancer. Scand. J. Clin. Lab. Invest. Suppl.. 106: 145-157. 1968.18. Fleckenstein. W.. and Weiss. C. Local tissue pO> measured with "thick" needle
probes. Funktionsanalyse Biolog. Systeme. 15: 155-166. 1986.19. Kallinowski. P., Zander. R.. Höckel,M., and Vaupel. P. Tumor tissue oxygénationas
evaluated by computeri/ed-pO^-histography. Int. J. Radial. Oncol. Biol. Phys., 19:953-961, 1990.
20. Vaupel, P., Schlcngcr. K.. Knoop, C., and Höckel,M. Oxygénationof human tumors:evaluation of tissue oxygen distribution in breastcancers by computeri/ed O> tensionmeasurements.Cancer Res.. 5/. 3316-3322. 1991.
21. Höckel.M.. Schienger. K., Knoop. C., and Vaupel. P. Oxygénationof carcinomas ofthe uterine cervix: evaluation by computerized O2 tension measurements. CancerRes., 51: 6098-6102. 1991.
22. Elias, H.. Henning. A., and Schwärt/,.D. E. Stereology: applications to biomédicalresearch. Hisiol. Rev.. SI: 158-200. 1971.
23. Weibel. H. R. Slereological principles for morphomelrv in electron microscopiccytology. Int. Rev. Cytol., 26: 235-302. 1969.
24. Höckel.M.. Knoop. C.. Schienger. K.. Vorndran. B.. Baussmann. E.. Mit/e. M..Knapstein. P. G., and Vaupel. P. Iniratumoral p()2 predicts survival in advancedcancer of the uterine cervix. Radiother. Oncol.. 26: 45-50, 1993.
25. Höckel,M., Vorndran. B.. Schienger. K.. Baussmann. li... Knapstein. P. G., andVaupel. P. Tumor oxygénation:a new predictive parameter in locally advancedcancer of the uterine cervix. Gynecol. Oncol.. 51: 141-149, 1993.
26. Höckel.M., and Knapstein. P. G. The combined operative and radiotherapeutictreatment (CORT) of recurrent tumors infiltrating the pelvic wall: first experiencewith 18 patients. Gynecol. Oncol.. 46: 20-28, 1992.
27. Stone. H. B., Brown. J. M.. Phillips. T. L., and Sutherland. R. M. Oxygen in humantumors: correlations between methods of measurement and response to therapy.Radial. Res.. 136: 422-434. 1993.
28. Nordsmark. M.. Bentzen. S. M., and Overgaard, J. Measurement of human tumouroxygénationstatusby a polarographic needleelectrode.Acta Oncol.. 33: 383-389. 1994.
29. Brizel, D. M.. Rosner. G. L.. Prosnitz, L. R., and Dewhirst. M. W. Patterns andvariability of tumor oxygénationin human soft tissue sarcomas,cervical carcinomas,and lymph node métastases.Int. J. Radial. Oncol. Biol. Phys.. 32: I 121-1 125. 1995.
30. Vaupel, P., Schaefer, C., and Okunieff. P. Intracellular acidosis in murine tlbrosar-comas coincides \vith ATP depletion, hypoxia. and high levels oÃláclateand total P,.NMR Biomed.. 7: 128-136. 1994.
31. Singer, S.. Souzu. K.. and Thiily. W. G. Pynivale utili/ation, phosphocholine andadenosine triphosphale (ATP) are markers ot human breast tumor progression: a -1IP-and '*C-nuclear magnetic resonance (NMR) spectroscopy study. Cancer Res.. 55:
5140-5145. 1995.32. Schwicken. G.. Walenta. S.. Sundfor. K.. Rofstad, E. K.. and M uel1er-Klieser. W.
Correlation of high láclalelevels in human cervical cancer with incidence of metastasis. Cancer Res., 55: 4757-4759. 1995.
33. Gray. L. H.. Conger, A. D.. Eben. M.. Hornsey. S.. and Scoti. O. C. A. Theconcentration of oxygen dissolved in tissues at the lime of irradiation as a factor inradiotherapy. Br. J. Radiol., 26: 638-648, 1953.
34. Horsnum. M. R.. Khali!. A.. Nordsmark. M., Grau, C., and Overgaard. J. Relationshipbetween radiobiological hypoxia and direct estimates of tumor oxygénationin amouse tumour model. Radiother. Oncol.. 2iS: 69-71. 1993.
35. Nordsmark. M.. Grau, C., Horsman. M. R.. Jörgensen.H. S., and Overgaard. J.Relationship between tumour oxygénation,hioenergetic status and radiobiologicalhypoxia in an experimental model. Acta Oncol., 34: 329-334, 1995.
36. Fenlon, B. M., Mohammad. F. K.. and Siemann. D. W. Should direct measurementsof tumor oxygénationrelate to the radiobiological Hypoxie fraction of a tumor'.' Int. J.
Radial. Oncol. Biol. Phys., 33: 365-373, 1995.37. Zywietz. F.. Reeker, W., and Kochs. E. Tumor oxygénationin a transplanted rat
rhabdomyosarcoma during fractionated irradiation. Int. J. Radiât.Oncol. Biol. Phys..32: 1391-1400. 1995.
38. Heacock, C. S.. and Sutherland. R. M. Induction characteristics of oxygen regulatedproteins. Int. J. Radial. Oncol. Biol. Phys., 12: 1287-1290. 1986.
39. Sutherland, R. M.. Ausserer. W., Murphy. B., and Laderoute. K. Tumor hypoxia andheterogeneity: challenges and opportunities for the future. Semin. Radial. Oncol.. 6:59-70. 1996.
40. Brizel. D. M.. Scully. S. P.. Harrelson. J. M.. Layfield. L. J.. Bean. J. M.. Prosnit/.L. R.. and Dewhirst. M. W. Tumor oxygénationpredicts the likelyhood ol distantmétastasesin human soft tissue sarcoma. Cancer Res.. 56: 941-943, 1996.
4515
on June 3, 2014. © 1996 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from
