polarographic electrode study of tumor oxygenation in clinically localized prostate cancer

doi:10.1016/S0360-3016(03)01621-3

CLINICAL INVESTIGATION Prostate

POLAROGRAPHIC ELECTRODE STUDY OF TUMOR OXYGENATION INCLINICALLY LOCALIZED PROSTATE CANCER

CHRIS PARKER, M.B.,*† MICHAEL MILOSEVIC, M.D.,*† ANTS TOI, M.D.,‡§ JOAN SWEET, M.D.,�¶

TONY PANZARELLA , M.SC.,# ROB BRISTOW, M.D., PH.D.,*† CHARLES CATTON, M.D.,*†

PAMELA CATTON, M.D.,*† JUANITA CROOK, M.D.,*† MARY GOSPODAROWICZ, M.D.,*†

MICHAEL MCLEAN, M.D.,*† PADRAIG WARDE, M.B.,*†AND RICHARD P. HILL , PH.D.**††

Departments of *Radiation Oncology,‡Medical Imaging,�Pathology,#Biostatistics, and** Experimental Therapeutics, PrincessMargaret Hospital, Toronto, ON, Canada; Departments of†Radiation Oncology,§Medical Imaging,¶Pathology, and

††Medical Biophysics, University of Toronto, Toronto, ON, Canada

Purpose: To describe the oxygenation of clinically localized prostate cancer.Methods and Materials: Intraprostatic oxygen tension was measured using the Eppendorf electrode in 55unanesthetized men with localized prostate cancer before radiotherapy. Measurements were made along twotracks through regions of suspected tumor in the prostate, and core needle biopsies were then obtained from thesame regions.Results: The median pO2 ranged from 0.2 to 57.3 mm Hg, and the grand median pO2 was 4.5 mm Hg. Thepercentage of oxygen readings <5 mm Hg (HP5) ranged from 0% to 100% (median 60%). The track 1 oxygenreadings were greater than those from track 2. Statistically significant heterogeneity was found in the individualoxygen readings: the between- and within-tumor components accounted for 32% and 68% of the total variability,respectively. However, the between-tumor variability in HP5 significantly exceeded the within-tumor variability(61% vs. 39%). No association was found between oxygen values and clinical factors, including age, T stage,Gleason score, prostate-specific antigen level, hemoglobin concentration, or prior hormonal treatment. Nodifference was noted in the oxygenation between regions of tumor and normal prostate tissue, as determined fromthe core biopsies.Conclusion: Localized prostate cancer is characterized by marked hypoxia and significant heterogeneity inoxygenation, similar to other human tumors. The normal prostate may contain regions of low oxygen concen-tration. HP5, as determined in this study, should adequately discriminate among patients with prostate cancerand allow the independent prognostic significance of oxygenation to be evaluated once the study matures.© 2004 Elsevier Inc.

Tumor hypoxia, Prostate neoplasms, Radiotherapy.

INTRODUCTION

Tumor hypoxia is a common feature of a range of humanmalignancies and has both prognostic and therapeutic sig-nificance. The degree of tumor hypoxia, as measured usingpolarographic electrodes, is an independent prognostic fac-tor for survival in both cervical carcinoma (1–3) and head-and-neck cancer (4). Randomized trials of strategies de-signed to improve tumor oxygenation or specifically targethypoxic cells have demonstrated survival benefits in pa-tients with non–small-cell lung cancer (5) and head-and-neck cancer (6, 7). The importance of hypoxia in prostatecancer is not known. However, low oxygen levels similar to

those observed in other tumor types have been described inanimal models of prostate cancer (8, 9) and in the only otherpolarographic electrode study to date of human prostateoxygenation (10, 11).

We devised a polarographic electrode technique to mea-sure the intraprostatic oxygen levels in awake, unanesthe-tized men with localized prostate cancer. Core needle biop-sies were obtained at the time of the measurements from thesame locations in the prostate to verify the presence orabsence of tumor. The ultimate objective of the study was toevaluate the independent prognostic significance of tumoroxygenation in patients with localized prostate cancer un-

Reprint requests to: Michael Milosevic, M.D., Department ofRadiation Oncology, Princess Margaret Hospital, 610 UniversityAve., Toronto, ON M5G 2M9 Canada. Tel: (416) 946-2124; Fax:(416) 946-2111; E-mail: [email protected]

Supported by the National Cancer Institute of Canada, thePrincess Margaret Hospital Foundation, and the Radiation Medi-

cine Program at the Princess Margaret Hospital.Acknowledgments—Thanks to Dr. Anthony Fyles for his helpfulsuggestions, Ami Syed for study coordination and data collection,and Hanxian Hu for data management.

Received Dec 12, 2002, and in revised form Jun 25, 2003.Accepted for publication Jul 25, 2003.

Int. J. Radiation Oncology Biol. Phys., Vol. 58, No. 3, pp. 750–757, 2004Copyright © 2004 Elsevier Inc.

Printed in the USA. All rights reserved0360-3016/04/$–see front matter

750

dergoing conformal radiotherapy (RT). The aims of thisreport were to describe the measurement technique; dem-onstrate that the technique is feasible and safe in a routineclinical setting and well tolerated by patients; describe theoxygenation of clinically localized prostate cancer; evaluatethe heterogeneity of oxygen readings in prostate cancer; andevaluate oxygenation in relation to the clinical characteris-tics.

METHODS AND MATERIALS

Patient selectionIntraprostatic measurements of oxygen tension were

made between June 1999 and July 2000 using the Eppen-dorf polarographic electrode system in men with clinicallylocalized prostate cancer before high-dose conformal RT.The eligibility criteria were a histologic diagnosis of pros-tatic adenocarcinoma, clinical Stage T1c, T2a-T2b, or T3a(1997 TNM classification), no evidence of lymph node ordistant metastases on CT scan of the abdomen/pelvis orbone scan, Eastern Cooperative Oncology Group perfor-mance status 0–2, age �80 years, a prior decision to treatwith high-dose conformal RT, and the ability to give in-formed consent.

Oxygen measurementsBefore receiving conformal RT for prostate cancer at our

institution, patients routinely have three inert gold seedsinserted into the prostate under transrectal ultrasound(TRUS) guidance. These seeds serve as fiducial markers forRT planning and verification purposes. Oxygen measure-ments were made immediately before seed insertion as partof the same procedure in awake, unanesthetized patientslying in the left lateral position. Prophylactic antibioticswere administered to minimize the risk of infection. All themeasurements were made by a single operator (A.T.) expe-rienced in the use of TRUS-guided biopsy of the prostate toensure technique consistency. The tumor location within theprostate gland was identified by palpation and TRUS, withthe benefit of previous biopsy information. Under directTRUS guidance, the tip of a custom-manufactured fine-needle Eppendorf electrode, 300 �m in diameter and 17 cmin length, was introduced transrectally and positioned in theposterior prostate just beneath the capsule until the oxygenreading stabilized (usually �1 min). Care was taken tominimize displacement or compression of the prostate glandwith insertion of the ultrasound probe. Between 20 and 25oxygen measurements were made 0.7 mm apart along alinear path through the tumor (track 1). The electrode wasthen retracted and a second series of 20–25 measurementswas obtained immediately along a separate linear trackthrough the tumor (track 2). The second measurement trackwas usually in the same region of the prostate as the first andseparated from the first by �1 cm. Movement of the needlethrough the tumor and the oxygen measurements along eachtrack were controlled automatically by the Eppendorf ma-chine, visualized on ultrasonography, and recorded on vid-

eotape. Oxygen values less than �3 mm Hg were assumedto be erroneous on the basis of previous calibration exper-iments and were deleted from the data set (12).

TRUS-guided core needle biopsies were obtained alongeach track after all the oxygen measurements were com-pleted and immediately before insertion of the marker seeds.The biopsies, which measured 1 mm in diameter and 10–20mm in length, were fixed in 10% neutral-buffered formalin,embedded in paraffin, and sectioned longitudinally at3–5-�m intervals. The sections were stained with hematox-ylin and eosin and examined using light microscopy by anexperienced genitourinary pathologist (J.S.) for the presenceor absence of carcinoma. In addition, discontinuous regionsof tumor and normal prostate tissue were mapped along thelength of the biopsy from superficial to deep for comparisonwith the corresponding oxygen measurements.

The Clinical Trials Committee of the Princess MargaretHospital and the Human Subjects Review Committee of theOffice of Research Services at the University of Torontoapproved the study. All patients provided informed writtenconsent. The measurements added approximately 5 min tothe duration of the routine marker seed insertion procedure.Patients were uncomfortable during the test because of theintrarectal ultrasound probe and experienced mild local painwith the measurements and biopsies. No complications re-sulted from the procedure, including no bleeding and noinfection.

Statistical analysisTumor oxygenation was described by the median pO2

and by the hypoxic proportion (HP5), defined as the per-centage of the 40–50 individual pO2 measurements in eachpatient (pooled results from tracks 1 and 2) that were �5mm Hg. The median pO2 and HP5 were also calculatedseparately for each track, and the results compared using theWilcoxon matched-pairs signed rank-sum test (13).

The intrinsic variability of individual oxygen measure-ments was estimated for comparison with other tumors(14–16). The total pO2 variance for the cohort (consideringall measurements in all patients) was divided into a be-tween-patient component and a within-patient component.The within-patient variance was divided again into a be-tween-track component and a within-track component. Thedata were analyzed using a nested linear mixed model (17).The track variable was nested within the patient variable.The between- and within-patient variances were expressedas percentages of the total (within-patient � between-pa-tient) variance.

The HP5 was used to evaluate the oxygen status inrelation to clinical prognostic factors. HP5 has been shownto predict the outcome in patients with uterine cervicalcancer (2) and was approximately normally distributed inthis population of prostate cancer patients. A total of 40individual oxygen measurements was predicted to yield aworst-case standard error in estimating HP5 of 7.9% (18).Doubling the number of individual measurements was pre-dicted to reduce the standard error only minimally to 5.6%.

751Prostate cancer oxygenation ● C. PARKER et al.

Therefore, measurements were made along only two tracksto provide acceptable accuracy and reproducibility whileminimizing the duration of patient discomfort and improv-ing the general tolerability and acceptance of the procedure.To verify the adequacy of this approach, the between-patient and within-patient variance components were esti-mated for HP5, assuming that each measurement track pro-vided an independent estimate of HP5 for a particular tumor.

The two-sample t test was used to compare oxygenationwhen two independent groups were defined (age dichoto-mized about the median value of 70 years, stage T1c vs.T2-T3a, Gleason score 6 vs. 7-8, prostatic volume dichot-omized about the median value of 48 cm3, hemoglobinconcentration dichotomized about the median value of 146g/L, neoadjuvant androgen deprivation [NAD] vs. no NAD,and positive vs. negative biopsy). One-way analysis ofvariance was used to compare the oxygenation in the threeclinical groups defined by a prostate-specific antigen (PSA)level �4.0 vs. 4.1–10 vs. �10 ng/mL.

All reported p values are two-tailed. All statistical anal-yses were performed using Statistical Analysis System/STAT software, version 8 (SAS Institute, Cary, NC).

RESULTS

Patient characteristicsProstate oxygenation was measured in 55 men with bi-

opsy-proven prostate cancer. The measurements were madebefore any treatment in 46 cases; the remaining 9 hadreceived NAD for a median duration of 3 months (range 3weeks to 8 months). The characteristics of the patients aresummarized in Table 1.

Prostate oxygenationA total of 40–50 individual oxygen readings was ob-

tained along two tracks in each of the 55 patients, yieldinga total of 2310 readings over 109 tracks (1 patient had 3tracks and 2 patients had only 1 track for technical reasons).The median pO2 in the 55 patients, taking all oxygen read-ings from both tracks into account, ranged from 0.2 to 57.3mm Hg, and the grand median pO2 (median of the 55median values) was 4.5 mm Hg. The HP5 ranged from 0%to 100% (median 60%). The distribution of these summarymeasures of oxygenation is illustrated in Fig. 1. A system-atic difference was found in the track 1 and track 2 oxygenreadings, with track 1 better oxygenated than track 2. Themean and median paired differences in the median pO2

(track 2 median pO2 � track 1 median pO2 in individualpatients) were �0.8 and �1.7 mm Hg, respectively (p �0.003), and the mean and median paired differences in HP5

were 14.4% and 4.8%, respectively (p � 0.008). The oxy-gen measurements are summarized in Table 2.

Heterogeneity of oxygen readingsThe evaluation of the 2310 individual oxygen readings

from both tracks in all 55 patients yielded a between-patientvariance of 32% and a within-patient variance of 68%. The

within-patient variance was further divided into a between-track component of 24% and a within-track component of44%. The between-track component was attributable to thesystematic difference in oxygenation between tracks 1 and 2that was described previously. No systematic difference wasobserved in the oxygen readings along the length of thetracks from superficial (subcapsular) to deep.

To validate the use of only two measurement tracks toevaluate prostate cancer oxygenation, the variance compo-nent analysis was repeated for HP5, assuming that eachmeasurement track in each tumor provided an independentestimate of HP5 for that tumor. There were 106 HP5 valuesdistributed among 53 tumors. The between-patient variancein HP5 comprised 62% of the total, and the within-patientvariance was 38%.

Oxygenation vs. clinical characteristicsThe pooled HP5, taking account of all oxygen readings

along both measurement tracks, was used to compare oxy-genation with the clinical characteristics (Table 3). Trendstoward greater hypoxia with advanced age, lower Gleasonscore, larger prostatic volume, and lack of prior NAD werenoted. However, none of these associations reached statis-tical significance.

Given the systematic difference in oxygen readings be-tween tracks 1 and 2, the relationship between HP5 and the

Table 1. Patient characteristics

Characteristic n*

Patients 55Age (y)

Median 70Range 52–79

Clinical T stage†

T1c 15T2a 34T2b 4T3a 2

Gleason score6 157–8 40

Initial PSA (ng/mL)�4 94.1–10 25�10 21

Prior NADYes 9No 46

Prostate volume‡ (cm3)Median 48Range 14–179

Hgb concentration (g/L)Median 146Range 115–187

Abbreviations: PSA � prostate-specific antigen; NAD � neo-adjuvant androgen deprivation; Hgb � hemoglobin.

* Number of patients, unless otherwise noted.† 1997 TNM classification.‡ From transrectal ultrasound.

752 I. J. Radiation Oncology ● Biology ● Physics Volume 58, Number 3, 2004

clinical characteristics of the patients was also analyzedseparately for each track. For track 1, the only factor thatwas associated with significantly lower oxygen readingswas the lack of prior NAD. The track 1 median HP5 was19% among the 9 patients who had received NAD and

57.1% among the remaining patients who had oxygen mea-surements performed before any treatment for prostate can-cer (p � 0.03). For track 2, no relationship was detectablebetween HP5 and any of the clinical variables, including theuse of NAD.

Fig. 1. (a) Distribution of median pO2 in 55 patients with localized prostate cancer (b) Percentage of oxygenmeasurements �5 mm Hg (HP5) in 55 patients with localized prostate cancer.


Oxygenation vs. pathologic findingsA core needle biopsy was obtained along each of the

Eppendorf electrode tracks immediately after the oxygen-ation measurements. The biopsies were 1 mm in diameterand ranged in length from 9.8 to 22.9 mm (median 14.2).Although the measurement and biopsy tracks were alignedas closely as possible using ultrasound guidance, it was notpossible to identify the Eppendorf track in the biopsy spec-imens. Both of the biopsies contained tumor in 34 cases,both were negative in 14 cases, and only one of the twobiopsies contained tumor in the remaining 7 cases. The

overall median HP5 (both tracks combined) was 69% for the41 cases with at least one positive biopsy and 50% whenboth biopsies were negative (p � 0.5). The track 1 medianHP5 was 61.9% for the biopsy-positive cases and 35.7% forthose that were biopsy-negative (p � 0.09). The corre-sponding track 2 median HP5 values were 71.4% and 69.0%(p � 0.87).

Discontinuous regions of tumor and normal prostate tis-sue were mapped along the length of the biopsy fromsuperficial to deep for comparison with the correspondingoxygen measurements. The median length of tumor andnormal prostate in each biopsy was 7.3 mm and 6.6 mm,respectively. No difference was found in the oxygenationbetween the tumor and normal prostate tissue (median HP5

62% and 59%, respectively). It was recognized that slightdifferences in the trajectory of the measurement and biopsyneedles, despite identical entry points, might lead to greaterspatial error at depth in the prostate. Furthermore, prostatecancer is usually located in the subcapsular peripheral zoneof the gland. For these reasons, the analysis was repeatedusing only the most peripheral 10 mm of tissue. Again, nodifference was observed between the oxygen readings intumor and normal prostate (median HP5 65% and 58%,respectively).

DISCUSSION

The results of this study confirmed that hypoxia of po-tential biologic significance exists in clinically localizedprostate cancer. The grand median pO2 of 4.5 mm Hg andthe median HP5 of 60% are similar to the previously re-ported results in cervical carcinoma and head-and-neck can-cer, tumor types in which oxygen level is an independentpredictor of survival (1–4). Most of our prostate tumorscontained hypoxic regions. The median pO2 values had anongaussian distribution (Fig. 1a), and most were clusteredin the narrow range of �10 mm Hg. This implies that evensmall differences in median pO2 may be indicative of po-tentially important differences in underlying tumor oxygen-ation. In contrast, HP5 was approximately normally distrib-uted between 0% and 100% and allowed greaterdiscrimination among tumors. We found no apparent rela-tionship between oxygenation and pretreatment clinicalprognostic factors (T stage, Gleason score, or PSA level).However, a trend toward higher oxygen readings was foundin patients who had received prior NAD. This requiresconfirmation in a larger cohort of patients with longerfollow-up.

Few other studies of human prostate cancer oxygenationhave been done, probably in part because of the relativeinaccessibility of the gland to needle electrodes. Rasey et al.(19) reported hypoxic proportions of between 0% and 94%in four human prostate cancers using positron emissiontomography of a fluorinated nitroimidazole compound thatis taken up in regions of hypoxia. The only other oxygenelectrode study in human prostate cancer was reported byMovsas et al. (10), and included 41 patients with clinically

Table 2. Summary of oxygen readings

Track

pO2 (mm Hg) HP5 (%)

Range Grand median Range Median

Overall (1 and 2) 0.2–57.3 4.5 0–100 601 1.5–75.4 4.9* 0–100 55†

2 0–70.4 3.0* 0–100 71†

Abbreviation: HP5 � hypoxic proportion (proportion of oxygenreadings �5 mm Hg).

* Mean paired difference �0.8 mm Hg, median paired differ-ence �1.7 mm Hg (p � 0.003).

† Mean paired difference 14.4%, median paired difference 4.8%(p � 0.008).

Table 3. Univariate analysis of HP5 vs. clinical characteristics

CharacteristicCases

(n)Overall median

HP5 (%)

Age (y)�70 (median) 32 56�70 23 69

T stageT1c 15 59T2–T3a 40 61

Gleason score6 15 697–8 40 56

PSA (ng/mL)0–4.0 9 504.1–10 25 64�10 21 54

Hgb concentration (g/L)�146 (median) 18 54�146 18 67Unknown 19 59

Prostate volume (cm3)�48 (median) 32 55�48 23 69

Prostate biopsyPositive 41 69Negative 14 50

Prior NADYes 9 48No 46 67

Abbreviations: HP5 � hypoxic proportion (proportion of oxy-gen readings �5 mm Hg); PSA � prostate-specific antigen; Hgb� hemoglobin; NAD � neoadjuvant androgen deprivation.


localized disease who were studied under spinal anestheticbefore prostate brachytherapy. Approximately 100 oxygenmeasurements along five needle tracks were obtained ineach patient from the involved portion of the prostate. Theoverall median pO2 was 6.3 mm Hg, similar to our result.Oxygenation was significantly better in patients �62 yearsold and in those with T1 vs. T2-T3 disease. A preliminaryoutcome analysis suggested that the ratio of prostate pO2 tonormal muscle pO2 predicted biochemical failure after treat-ment (20).

The measurement of prostate cancer oxygenation beforeexternal beam RT is technically difficult because of theanatomic position of the prostate gland and necessitates acompromise between the two fundamental goals of obtain-ing reliable and reproducible information about tumor ox-ygenation and, at the same time, minimizing patient dis-comfort and inconvenience. Our approach addressed theseissues by coupling the oxygen measurements with the rou-tine procedure of fiducial marker insertion that all ourpatients with prostate cancer undergo before conformal RT.This allowed transrectal access to the prostate gland foroxygen measurements without the need for an additionalhospital visit or additional patient preparation (laxatives,prophylactic antibiotics). Current TRUS and biopsy tech-niques allow the measurements and extra biopsies to beobtained in awake, unanesthetized patients with minimaldiscomfort and a low risk of side effects. The combinationof previous diagnostic biopsy information, palpation of theprostate gland, ultrasound echogenicity, and Doppler bloodflow was used to target regions of tumor and was accuratein 75% of cases as determined by biopsies obtained from thesame regions immediately after the measurements.

The Eppendorf electrode measures the average oxygenconcentration in a volume of tissue at the tip of the electrodethat might contain tumor cells, tumor interstitium, bloodvessels, and normal prostate tissue. In addition, both spatialand time-dependent variations exist in the oxygen concen-tration. These factors contribute to the variability in oxy-genation from region to region. The within-tumor variabil-ity of individual oxygen readings in prostate canceraccounted for 68% of the total sample variability. In com-parison, Nordsmark et al. (15) described a within-tumorvariance component of 75% in head-and-neck cancer and69% in soft-tissue sarcoma. Wong et al. (16) reported thewithin-tumor component in cervical cancer to be 67% of thetotal variance. These results imply remarkable similarity inthe intrinsic heterogeneity of individual oxygen readings intumors of different types, despite variations in histologictypes and patterns of growth. They also underscore theimportance of obtaining multiple measurements in eachtumor to evaluate overall oxygenation reliably and repro-ducibly. Repeat measurements done in the same manner atlater times should yield the same result within the accuracylimits of the test, assuming no change in the true underlyingoxygen status of the tumor. It can be estimated that 40individual oxygen readings in each tumor would produce astandard error in the estimation of HP5 of 7.9% and that

increasing the number of measurements would reduce thestandard error in a diminishing fashion according to aninverse square-root function (18). Therefore, 40 measure-ments were chosen in this study as a compromise betweenreproducibility and patient acceptance of the procedure. Toassess the validity of this approach, the variance-componentanalysis was repeated for HP5, assuming that each trackprovided an independent estimate of HP5 for a particulartumor. The between-tumor component accounted for 61%of the total sample variance in HP5 and significantly ex-ceeded the within-tumor component. Therefore, HP5, asmeasured in this study, should adequately discriminateamong patients for the purpose of evaluating the indepen-dent prognostic significance of oxygenation in prostate can-cer.

We found the oxygen readings to be influenced by thesequence in which the measurements were made. The track1 pO2 readings were significantly greater than the track 2readings, although no systematic differences were noted inoxygenation along the length of the measurement tracks.This is in contrast to our experience in cervical cancer, inwhich no systematic differences were observed among mea-surements from different tracks in the same tumor (16). Itraises concern that the process of making measurementsalong the first track might have influenced the results ob-tained from the second track. Possible explanations includepressure from the intrarectal ultrasound probe, vasculardisruption, reactive vasoconstriction, bleeding, or edema.Although this observation implies the need for caution whenpooling oxygen measurements from multiple tracks in pa-tients with prostate cancer, the magnitude of the differencebetween tracks was small (mean paired difference in HP5

14.4%, median paired difference 4.8%) and therefore prob-ably of minimal clinical significance. Movsas et al. (10, 11)did not specifically comment on differences in the oxygenreadings among measurement tracks in their patients withprostate cancer.

A trend was found toward lower oxygen readings if theaccompanying core needle biopsies from the region of themeasurements contained tumor. However, the relationshipfailed to achieve statistical significance. Early prostate can-cer is characterized by infiltrative, multifocal growth, anddiscontinuous regions of tumor and normal prostate werefrequently seen in the biopsies. To evaluate more accuratelythe oxygenation of prostate cancer as distinct from normalprostate tissue, the location of tumor and normal prostatewas mapped along the length of each biopsy and correlatedwith the corresponding oxygen readings. No difference inoxygenation was detected between tumor and normal tissue(median HP5 62% and 59%, respectively). The measure-ment tracks and biopsies were closely aligned using ultra-sound guidance, but it was not possible to identify theelectrode track reliably in the biopsy specimens. It wasrecognized that slight differences in the trajectory of themeasurement and biopsy needles, despite identical entrypoints, might lead to greater spatial error at depth in theprostate. Therefore, the mapping analysis was repeated us-


ing only the most peripheral 10 mm of the prostate biopsies.This is likely to encompass most of the tumor, which isusually located in the subcapsular peripheral zone of thegland. Despite this more rigorous analysis, no differencewas found in the oxygenation between the regions. Theseresults suggest that the normal prostate gland may containregions that are poorly oxygenated and comparable in thisrespect to tumor. This will need to be addressed in futurestudies. An important question is whether “contamination”of the tumor oxygen readings by normal tissue measure-ments, although similar in value, might have influenced theprognostic value of the procedure. Movsas et al. (20), in apreliminary analysis, described an association between ox-ygen measurements made randomly in the involved regionof the prostate (without pathologic confirmation of tumor atthe measurement site) and outcome after RT, suggestingthat more precise localization of the electrode in tumor maynot be necessary.

Evidence is mounting that hypoxia exists in prostatecancer, as it does in other human tumors. Hypoxia promotesa range of genetic alterations implicated in malignant pro-

gression (21), including the selection of cells with mutationsof the p53 tumor suppressor gene (22). In addition, hypoxiais an important stimulus of angiogenesis, and several studieshave shown increased expression of vascular endothelialgrowth factor and other angiogenic proteins in prostatetumors (23–25). A correlation between needle electrodeoxygen measurements and vascular endothelial growth fac-tor expression has been demonstrated in patients with pros-tate cancer before prostatectomy (26). Both p53 mutationsand high levels of angiogenesis have been linked to earlytumor progression after RT for prostate cancer and loweroverall patient survival (27–29). Hormonal therapy of an-drogen-sensitive prostate cancer may inhibit angiogenesis(30, 31) and lead to improved tumor oxygenation (32).These observations, together with preliminary needle elec-trode results (20), support the hypothesis that hypoxia is animportant cause of RT failure in prostate cancer. Our studycontinues to accrue and mature and will eventually allow usto assess the interaction between hormonal treatment andoxygenation, as well as the independent prognostic signifi-cance of oxygen measurements in this disease.

REFERENCES

1. Hockel M, Knoop C, Schlenger K, et al. Intratumoral pO2

predicts survival in advanced cancer of the uterine cervix.Radiother Oncol 1993;26:45–50.

2. Fyles A, Milosevic M, Hedley D, et al. Tumor hypoxia hasindependent predictor impact only in patients with node-neg-ative cervix cancer. J Clin Oncol 2002;20:610–615.

3. Rofstad EK, Sundfor K, Lyng H, et al. Hypoxia-inducedtreatment failure in advanced squamous cell carcinoma of theuterine cervix is primarily due to hypoxia-induced radiationresistance rather than hypoxia-induced metastasis. Br J Can-cer 2000;83:354–359.

4. Brizel DM, Dodge RK, Clough RW, et al. Oxygenation ofhead and neck cancer: Changes during radiotherapy and im-pact on treatment outcome. Radiother Oncol 1999;53:113–117.

5. von Pawel J, von Roemeling R, Gatzemeier U, et al. Tira-pazamine plus cisplatin versus cisplatin in advanced non-small-cell lung cancer: A report of the international CATA-PULT I (Cisplatin and Tirapazamine in Subjects withAdvanced Previously Untreated Non-Small-Cell Lung Tu-mors) study group. J Clin Oncol 2000;18:1351–1359.

6. Haffty BG, Hurley R, Peters LJ. Radiation therapy with hy-perbaric oxygen at 4 atmospheres pressure in the managementof squamous cell carcinoma of the head and neck: Results ofa randomized clinical trial. Cancer J Sci Am 1999;5:341–347.

7. Overgaard J, Hansen HS, Overgaard M, et al. A randomizeddouble-blind phase III study of nimorazole as a hypoxic ra-diosensitizer of primary radiotherapy in supraglottic larynxand pharynx carcinoma: Results of the Danish Head and NeckCancer Study (DAHANCA) protocol 5-8. Radiother Oncol1998;46:135–146.

8. Leith JT, Quaranto L, Padfield G, et al. Radiobiological stud-ies of PC-3 and DU-145 human prostate cancer cells: X-raysensitivity in vitro and hypoxic fractions of xenografted tu-mors in vivo. Int J Radiat Oncol Biol Phys 1993;25:283–287.

9. Yeh KA, Biade S, Lanciano RM, et al. Polarographic needleelectrode measurements of oxygen in rat prostate carcinomas:Accuracy and reproducibility. Int J Radiat Oncol Biol Phys1995;33:111–118.

10. Movsas B, Chapman JD, Greenberg RE, et al. Increasinglevels of hypoxia in prostate carcinoma correlate significantlywith increasing clinical stage and patient age. Cancer 2000;89:2018–2024.

11. Movsas B, Chapman JD, Horwitz EM, et al. Hypoxic regionsexist in human prostate carcinoma. Urology 1999;53:11–18.

12. Kavanagh MC, Sun A, Hu Q, et al. Comparing techniques ofmeasuring tumor hypoxia in different murine tumors: Eppen-dorf pO2 histograph [3H]misonidazole binding and pairedsurvival assay. Radiat Res 1996;145:491–500.

13. Altman DG. Practical statistics for medical research. London:Chapman and Hall; 1991.

14. Brizel DM, Rosner GL, Prosnitz LR, et al. Patterns andvariability of tumor oxygenation in human soft tissue sarco-mas, cervical carcinomas, and lymph node metastases. Int JRadiat Oncol Biol Phys 1995;32:1121–1125.

15. Nordsmark M, Bentzen SM, Overgaard J. Measurement ofhuman tumour oxygenation status by a polarographic needleelectrode: An analysis of inter- and intratumour heterogeneity.Acta Oncol 1994;33:383–389.

16. Wong RK, Fyles A, Milosevic M, et al. Heterogeneity ofpolarographic oxygen tension measurements in cervix cancer:An evaluation of within and between tumor variability, probeposition, and track depth. Int J Radiat Oncol Biol Phys 1997;39:405–412.

17. Littell RC, Milliken GA, Stroup WW, et al. SAS for mixedmodels. Cary, NC: SAS Institute Inc.; 1996.

18. Fleiss JL. Statistical methods for rates and proportions. NewYork: John Wiley & Sons; 1981.

19. Rasey JS, Koh WJ, Evans ML, et al. Quantifying regionalhypoxia in human tumors with positron emission tomographyof [18F] fluoromisonidazole: A pretherapy study of 37 pa-tients. Int J Radiat Oncol Biol Phys 1996;36:417–428.

20. Movsas B, Chapman JD, Hanlon AL, et al. A hypoxic ratio ofprostate pO2/muscle pO2 predicts for biochemical failure inprostate cancer patients. Int J Radiat Oncol Biol Phys 2001;51(Suppl. 1):111.

21. Dachs GU, Chaplin DJ. Microenvironmental control of gene


expression: Implications for tumor angiogenesis, progression,and metastasis. Semin Radiat Oncol 1998;8:208–216.

22. Graeber TG, Osmanian C, Jacks T, et al. Hypoxia-mediatedselection of cells with diminished apoptotic potential in solidtumours. Nature 1996;379:88–91.

23. Ferrer FF, Miller LJ, Andrawis RI, et al. Vascular endothelialgrowth factor (VEGF) expression in human prostate cancer: Insitu and in vitro expression of VEGF by human prostatecancer cells. J Urol 1997;157:2329–2333.

24. Ferrer FF, Miller LJ, Andrawis RI, et al. Angiogenesis andprostate cancer: In vivo and in vitro expression of angiogen-esis factors by prostate cancer cells. Urology 1998;51:161–167.

25. Jackson MW, Bentel JM, Tilley WD. Vascular endothelialgrowth factor (VEGF) expression in prostate cancer and be-nign prostatic epithelium. J Urol 1997;157:2323–2328.

26. Cvetkovic D, Movsas B, Dicker AP, et al. Increased hypoxiacorrelates with increased expression of the angiogenesismarker vascular endothelial growth factor in human prostatecancer. Urology 2001;57:821–825.

27. Hall MC, Troncoso P, Pollack A, et al. Significance of tumor

angiogenesis in clinically localized prostate carcinoma treatedwith external beam radiotherapy. Urology 1994;44:869–875.

28. Osman I, Drobnjak M, Fazzari M, et al. Inactivation of the p53pathway in prostate cancer: Impact on tumor progression. ClinCancer Res 1999;5:2082–2088.

29. Silberman MA, Partin AW, Veltri RW, et al. Tumor angio-genesis correlates with progression after radical prostatectomybut not with pathologic stage in Gleason sum 5 to 7 adeno-carcinoma of the prostate. Cancer 1997;79:772–779.

30. Levine AC, Liu XH, Greenberg PD, et al. Androgens inducethe expression of vascular endothelial growth factor in humanfetal prostatic fibroblasts. Endocrinology 1998;139:4672–4678.

31. Hartley-Asp B, Vukanovic J, Joseph IB, et al. Anti-angiogenictreatment with linomide as adjuvant to surgical castration inexperimental prostate cancer. J Urol 1997;158:902–907.

32. Hansen-Algenstaedt N, Stoll BR, Padera TP, et al. Tumoroxygenation in hormone-dependent tumors during vascularendothelial growth factor receptor-2 blockade, hormone abla-tion, and chemotherapy. Cancer Res 2000;60:4556–4560.


polarographic electrode study of tumor oxygenation in clinically localized prostate cancer

Documents