Comparison of Serum PSMA, PSA Levels WithResults of Cytogen-356 ProstaScintT Scanning in

Prostatic Cancer PatientsGerald P. Murphy,1* Robert T. Maguire,2 Barbara Rogers,3 Alan W. Partin,4

Wil B. Nelp,5 Michael J. Troychak,6 Haakon Ragde,6 Gerald M. Kenny,6Robert J. Barren III,1 Victoria A. Bowes,1 A.K. Gregorakis,1 Eric H. Holmes,7

and Alton L. Boynton8

1Pacific Northwest Cancer Foundation/Northwest Hospital, Seattle, Washington2Targon Corporation, Princeton, New Jersey3Cytogen Corporation, Princeton, New Jersey

4Department of Urology, The Johns Hopkins Hospital, Baltimore, Maryland5Department of Nuclear Medicine, University of Washington, Seattle, Washington

6Northwest Hospital, Seattle, Washington7Pacific Northwest Cancer Foundation, Cancer Research Division, Department of Cell

Surface Biochemistry, Northwest Hospital, Seattle, Washington8Pacific Northwest Cancer Foundation, Cancer Research Division, Department of Molecular

Medicine, Northwest Hospital, Seattle, Washington

BACKGROUND. Stored serum from clinical trial cases undergoing ProstaScintt (CYT-356)scanning were available for Prostate Specific Membrane Antigen (PSMA) assay. ProstateSpecific Antigen (PSA) levels had already been determined. This provided an opportunity tosee what correlations existed between the serum markers and the ProstaScintt scan. A groupof patients had the studies preprostatectomy, whereas another group had the studies post-prostatectomy.METHODS. The scan results, serum PSA, serum PSMA, and clinical data were separatelyanalyzed. PSMA serum levels were determined by Western blot.RESULTS. Preoperatively, radical prostatectomy patients showed a correlation between se-rum PSA or PSMA levels and the ProstaScintt scan in the total group (n = 86), or in anuntreated group (n = 38). Preoperatively, PSMA correlated with the pathological stage,whereas PSA correlated with the scan. Postoperatively, only PSMA serum levels correlatedwith the scan in an untreated group (n = 40).CONCLUSIONS. Preoperatively or postoperatively, Western blot PSMA serum levels predictthe stage of disease or local, regional, or distant metastases, as shown by ProstaScintt scan.Both the scan and the serum tests provide prognostic information and evaluate the extent ofdisease to a more significant degree than previously possible. Prostate 33:281–285, 1997.© 1997 Wiley-Liss, Inc.

KEY WORDS: prostate cancer; ProstaScintt scan; PSMA; prostate-specific membraneantigen; PSA; prostate-specific antigen; pathologic stage

INTRODUCTION

The identification of a monoclonal antibody (7E11-C5.3) reactive with prostate cancer cells has led to theidentification of an antigen, PSMA (prostate-specificmembrane antigen). When radiolabeled with indium,

A.K. Gregorakis is presently at the Department of Urology, Evan-gelismos Hospital, Athens, Greece.*Correspondence to: Gerald P. Murphy, Pacific Northwest CancerFoundation/Northwest Hospital, 120 Northgate Plaza, Suite 205,Seattle, WA 98125-7001.Received 7 July 1997; Accepted 14 July 1997

The Prostate 33:281–285 (1997)

© 1997 Wiley-Liss, Inc.

the antigen (ProstaScintt, CYT-356, Cytogen Corpora-tion, Princeton, NJ) can be used to detect occult pros-tate cancer by single photon emission computerizedtomography (SPECT) scanning [1–9]. Recent effortshave compared the serum levels of PSMA with total orfree-PSA (prostate-specific antigen), emphasizing theprognostic value of PSMA [10]. However, there havebeen no comparisons of the imaging with ProstaScinttserum levels of PSMA or PSA. This is the basis for thisreport.

MATERIALS AND METHODS

Patient Populations

All patients within this study were part of clinicalstudies sponsored by the Cytogen Corporation. Thepurpose of these studies was to evaluate the utility ofthe radiolabeled monoclonal antibody CYT 356 for im-aging of distant and local prostate cancer metastases[8,9,11,12,13]. The ProstaScintt scans were performedat multiple centers. All scans were separately evalu-ated by a participating consulting nuclear medicinephysician. Blood sample markers were obtained fromthe patients within 30 days of the ProstaScintt proce-dure.

The patients in this study consisted of those pa-tients (group A) with intact prostates (86 patients) orgroup B, who had undergone a previous radical pros-tatectomy (150 patients). A total of 236 patients wasavailable for evaluation. The groups were further sub-divided into those patients not receiving any treat-ment or others who received one of three treatments:hormone ablation, radiation, or chemotherapy. Of thepostprostatectomy group, 40 patients were not on anytreatment and 110 had received therapy. There were38 untreated patients in the preradical group A, andan additional 48 in the preradical group who receivedtherapy.

ProstaScintT Scan

7E11-C5.3 (CYT-356) is a murine monoclonal anti-body, prepared against membrane fractions of LNCaPcells [1]. The antigen targeted by this antibody wasPSMA. PSMA has been shown to have homologiesto the transferrin receptor and the brain proteinNAALADase [12,13]. Through histological studies onparaffin-embedded tissue, a 7E11-C5 specificity forprostate epithelial tissue has been shown [1]. There is,however, weak expression also in salivary ducts andthe lower intestine, as well as in smooth muscle [3,13].

In preparation for the ProstaScintt infusion, 7E11-C5.3 was covalently modified with glycyl-tyrosyl-(N-E-diethylenetriaminepentaacetic acid)-lysine hydro-chloride (7E11-C5.3-GYK-DTPA) through oxidation of

the oligosaccharide moieties on the Fc region of theantibody [8,9]. A total of 0.5 mg of 7E11-C5.3-GYK-DTPA was labeled with 5 mCi of 111indium chlorideprior to infusion. Infusions were given in a slow in-travenous push and patients were monitored at 5, 30,and 60 min after infusion [8,9].

Imaging was done between 2–5 days postinfusion.Planar images of the anterior and posterior thorax,and abdomen/pelvis, were done using a large field-of-view gamma camera equipped with a medium en-ergy collimator. Images of the proximal humeri, fe-murs, and skull were acquired with a second imagingsession. Imaging of the abdomen/pelvis were doneusing SPECT. Reconstruction of raw data was doneusing a Weiner filter on the computer [8]. All imageswere reviewed in the transverse, coronal, and sagittalplanes. All scans were reviewed by a nuclear medicinephysician.

Scan results were reported as localized, regional, ordistant, or in some cases, in combination. Localizedpositive cases were confined to the prostate bed, pros-tatic fossa, or seminal vesicles. In cases where patientsstill had their prostates intact (group A), local scanresults were not considered to be positive for purposesof statistical analysis. Regional positive results weredefined as any mass located in the pelvis, pelvic wall,or extension of the tumor to the following lymphnodes: iliac, presacral, inguinal, hypogastric, obtura-tor, pelvic periprostatic, common iliac, external iliac,internal iliac, iliac bifurcation, and perirectal. Distantpositive results included all cases with a positive scanoutside the pelvis.

For the purpose of receiver-operator analysis, posi-tive and negative populations were defined as follows:in postprostatectomy patient groups, a positive scanwas any positive result, local, regional, or distant. Incases where patients had an intact prostate, a localpositive scan result was placed in the ‘‘negative’’population, as stated previously. Regional or distantpositive scans were grouped as positive. For purposesof correlation in the preprostatectomy group, popula-tions were defined as follows: group 1, negative (ex-cluding local scan results); group 2, regionally posi-tive, or regional only; group 3, distant, or distant only;and group 4, all regions positive.

Staging of Patients

Pathological staging was completed at each report-ing center by the pathologist using the TNM system[14]. Clinical staging was completed by the attendingphysician.

PSA Assay

Assays for prostate-specific antigen were per-formed at the centers where the ProstaScintt scans

282 Murphy et al.

were performed. The Hybritech Tandem-R PSA assayfor prostate-specific antigen (Hybritech Incorporated,San Diego, CA), a two-site radioactive sandwich assayfor PSA, was used for this purpose. Assays for pros-tate-specific antigen and prostate-specific membraneantigen were performed on samples that were drawnon the same day.

PSMA Western Blot

Serum samples were diluted 1:7 in lysis buffer (1%Triton X-100, 50 mM HEPES, 10% glycerol, 15 mMMgCl2, 1 mM AEBSF, 1 mM EGTA). The diluted se-rum samples were then combined at a ratio of 2:3 withsample buffer (SDS reducing buffer). Samples wererun on 8.5% SDS-polyacrylamide gel electrophoresis(final protein concentration of 0.44 mg/ml per sample,as determined using the Bio-Rad Protein Assay, Bio-Rad Laboratories, Hercules, CA). The separated pro-teins were transferred onto PVDF membrane using theBio-Rad Mini Trans-Blot Electrophoretic Transfer Cellat 90 v for 1 hr at 4°C (Bio-Rad Laboratories) in trans-fer buffer (Tris, glycine, methanol). Membranes werethen probed with 3 mg/ml 7E11.C5 antibody, and 167ng/ml sheep anti-mouse HRP-labeled secondary an-tibody, and the membrane was developed using theChemiluminescent Substrate Kit (Kirkegaard & PerryLaboratories, Inc., Gaithersburg, MD).

Blots were visualized by exposing X-ray film, re-vealing a protein band of approximately 120 kD. Theblot image was scanned with a Microtek ScanMakerIIHR scanner (Microtek Labs, Incorporated, RedondoBeach, CA), and band intensities were measured by ananalysis performed on a Power Macintosh 7100/80computer (Apple Computer, Cupertino, CA) using thepublic domain NIH Image program (written byWayne Rasband at the U.S. National Institutes ofHealth and available from the Internet by anonymousftp from zippy.nimh.nih.gov or on floppy disk fromNTIS, 5285 Port Royal Rd., Springfield, VA 22161, partnumber PB93-504868). All patient samples were as-sessed against known low and high PSMA valuesfrom the same Western blot as standard controls[10,15].

The current serum normal range for PSMA inhealthy individuals (age 35–55 years) without knownprostate cancer is 0.1384–0.2198 (relative intensity lev-els). The control range is based on over 300 assaysperformed on multiple occasions from a pool of do-nors who were evaluated by general health examina-tions, including prostate examinations (by PSA assay),and who had no known history of prostate disease orcancer.

Statistics

The populations were analyzed using Microsoft Ex-cel (Microsoft, Inc., Redmond, WA) and Systat (SPSS,Inc., Chicago, IL) [16]. Receiver operator curves weregenerated based on the sensitivity and specificity ofthe data sets, using positive and negative ProstaScinttscans [17,18]. Statistical analysis was done using Sys-tat. Least squares means, correlation, correlation P val-ues and Kruskal-Wallis tests were calculated by Systat[16]. The Kruskal-Wallis test was used to test the vari-ance of the markers across pathological stage, scans,and Gleason score [19]. For correlation and Kruskal-Wallis tests, P < 0.05 was viewed as significant.

RESULTS

Table I illustrates the correlation seen in the preop-erative (radical prostatectomy) cases. PSA correlatedwith the scan (P = 0.014), while PSMA serum valuessignificantly correlated with the pathologic stage ofthe operative specimen (P = 0.006). The correlation ofPSMA with the stage or scan in the preoperativegroup was noted in the total group (n = 86), or in thepreoperative untreated group (n = 38) (P ø .05), butnot the preoperative treated group (n = 48). Bone scanswhen performed in these patients were negative.

Table II depicts the mean values for PSMA or PSAby stage, showing a correlation only for PSMA serumlevels.

In the total postprostatectomy group (n = 150), nosignificant correlations were noted in the treatedgroup (n = 110). Bone scans were negative in all pa-tients. Table III illustrates the significant (P = 0.017)correlation observed only between the serum PSMAvalues and the ProstaScintt scan in the postprostatec-tomy untreated patients (n = 40).

Table IV demonstrates that the positive scans re-lated to the serum PSMA levels in terms of overalltumor burden, and location, e.g., greater for distantlesions compared to local recurrence (P = 0.017).

TABLE I. Correlations Between Serum PSA, PSMA,and ProstaScintT Scan in Preoperative Patients*

Patients(n = 86) ROCa area

Correlations

Gleason Stage Scan

PSA 0.7083 NS NS r2 = 0.146P = 0.014

PSMA 0.5421 NS r2 = 0.144 NSP = 0.006

*Normal range, 0.080–0.220 units; PSA normal range, 0–4.0 ng/ml.aROC, Receiver Operating Characteristic Curve.

Correlation of Serum PSMA and Prostate Scans 283

DISCUSSION

Table V is a summary relating serum PSMA or PSAlevels to the various preoperative or postoperativegroups with or without treatment. There was no otherevidence of metastatic disease as judged by bonescans. PSMA correlates best comparing stage of theresected prostate to the preoperative scan. However,treatment preoperatively eliminated that correlation.The Gleason score showed no correlation at any timein this series. In results already reported from the Cy-togen clinical studies by Nelp, the ProstaScintt scanin preoperative cases had a sensitivity of 62%, speci-ficity of 72%, and 68% accuracy when compared to theresults of pelvic lymph node dissections [20]. Inpostradical prostatectomy cases, ProstaScintt had asensitivity of 49%, specificity of 71%, and accuracy of63% compared to fossa biopsy [20]. In a recent studyutilizing Tc-99m-labeled 7E11.C5 antibody (Tc-99m-labeled CYT-351), an overall accuracy of 92% wasfound [21]. Thus, a change to Tc-99m may make iteasier to read the scans, as well as to perform the testmore quickly. The Western blot assay for PSMA serumlevels is also an interim step. Additional monoclonalantibodies have been prepared, and a classical sand-wich assay will be available soon. At this point evenbetter correlations should be noted between serum

PSMA and stage of the primary tumor and the scan.PSMA reflects the presence of clonal-resistant cells, aspostulated originally, and thus will always be of betterclinical prognostic value compared to PSA [1]. Prosta-Scintt scans provide a new opportunity to evaluatepatients prior to or after treatment for occult disease insoft tissue or lymph nodes, previously not a possibil-ity. Serum PSMA levels will likely assist in selectingpatients for immunoscintigraphy. ProstaScintt scan-ning offers an opportunity to further determine thepresence of lymph node metastases. A patient positivefor metastatic disease will likely be treated differentlythan one believed to have localized, curable disease[22,23]. The discovery of the 7E11.C5.3 antibody hasmade this possible [1]. Pelvic lymphadenectomy is stilla sampling procedure and may miss occult cancers[22]. Noninvasive modalities, including ultrasound,CAT scans, and MRI, have been far less than ideal intheir performance [24,25]. ProstaScintt scanning asperformed in the studies reported here raises this per-formance level considerably. Further, PSMA serumlevels appear to correlate well with a positive scan andare relatively independent of total PSA values as re-corded. Further improvements, both in the serumPSMA assay and in scan acquisition and interpreta-tion, will doubtless be of great value.

ACKNOWLEDGMENTS

We acknowledge support from the Phi Beta Psi So-rority (Crosslake, MN); the CaP CURE Foundation(Santa Monica, CA); Hybritech, Incorporated, a sub-sidiary of Beckman Instruments, Inc. (San Diego, CA);the Ed Rontell Advisory Fund (Fresno, CA); and Dr.Graham May, Vice President for Medical Affairs, Cy-togen Corporation (Princeton, NJ), who provided in-put on the manuscript during preparation.

TABLE II. Pathological Stage of ReconstructedProstates Compared to Serum PSA or PSMA Levels, in

the Total Population*

Stage PSMA ±SE PSA ±SE n

pT1a 0.144 0.043 39.344 24.567 7pT2a 0.273 0.022 60.290 12.070 23pT2b 0.245 0.056 38.750 32.499 2pT3a 0.288 0.015 57.988 9.319 52pT3c 0.394 0.078 75.100 47.168 2

r2 = 0.144 NSP = 0.006 NS

*n, number; SE, standard error; p, pathologic stage.

TABLE III. Correlation Between ProstaScintT Scan andSerum PSMA Observed in Postradical

Untreated Patients*

Patients(n = 40) ROC area

Correlations

Gleason Stage Scan

PSA 0.4487 NS NS NS

PSMA 0.5266 NS NSr2 = 0.308P = 0.017

*Average follow-up time, 2.453 years (896 days).

TABLE IV. Specific Site for ProstaScintT ScanCorrelated With Serum Markers In Postradical

Prostatectomy, Untreated Patients*

Scan PSMA ±SE PSA ±SE n

Negative 0.287 0.020 12.965 7.518 19Local 0.173 0.049 2.333 3.172 3Regional 0.284 0.050 2.467 19.411 3Local regional 0.310 0.050 2.50 3.885 2Distant 0.375 0.032 7.629 2.076 7Local distant 0.288 0.061 2.650 23.774 2Regional distant 0.323 0.050 9.267 19.411 3Local regional distant 0.071 1.800

r2 = 0308 NSP = 0.017 NS

*n, number; SE, standard error.

284 Murphy et al.

REFERENCES

1. Horoszewicz JS, Kawinski E, Murphy GP: Monoclonal antibod-ies to a new antigenic marker in epithelial prostatic cells andserum of prostatic cancer patients. Anticancer Res 1987;7:927–936.

2. Wynant GE, Murphy GP, Horoszewicz JS, Neal CE, Collier BD,Mitchell E, Purnell G, Tyson I, Abdel-Nabi H, Winzelberg G:Immunoscintigraphy of prostatic cancer: Preliminary resultswith 111In-labeled monoclonal antibody 7E11-C5.3 (CYT-356).Prostate 1991;18:229–241.

3. Murphy GP: Radioscintiscanning of prostate cancer. Cancer1995;75:1819–1822.

4. Rochon YP, Horoszewicz JS, Boynton AL, Holmes EH, BarrenRJ III, Erickson SJ, Kenny GM, Murphy GP: Western blot assayfor prostate-specific membrane antigen in serum of prostatecancer patients. Prostate 1994;25:219–223.

5. Murphy GP, Holmes EH, Boynton AL, Kenny GM, OstensonRC, Erickson SJ, Barren RJ: A comparison of prostate specificantigen, prostate specific membrane antigen, and LNCaP basedELISA assays in prostatic cancer patients and patients with be-nign prostatic enlargement. Prostate 1995;26:164–168.

6. Murphy G, Ragde H, Kenny G, Barren R III, Erickson S, Tjoa B,Boynton A, Holmes E, Gilbaugh J, Douglas T: Comparison ofprostate specific membrane antigen, and prostate specific anti-gen levels in prostatic cancer patients. Anticancer Res 1995;15:1473–1480.

7. Maguire RT, Pascucci VL, Maroli AN, Gulfo JV: Immunoscin-tigraphy in patients with colorectal, ovarian, and prostate can-cer. Results with site specific immunoconjugates. Cancer[Suppl] 1993;72:3453–3462.

8. Babaian RJ, Sayer J, Podoloff DA, Steelhammer LC, Bhadkam-kar VA, Gulfo JV: Radioimmunoscintigraphy of pelvic lymphnodes with Indium 111-labeled monoclonal antibody CYT-356.J Urol 1994;152:1952–1955.

9. Kahn D, Williams RD, Seldin DW, Libertino JA, Hirschhorn M,Dreicer R, Weiner GJ, Bushnell D, Gulfo J: Radioimmunoscin-tigraphy with 111Indium labeled CYT-356 for the detection ofoccult prostate cancer recurrence. J Urol 1994;152:1490–1495.

10. Murphy GP, Barren RJ, Erickson SJ, Bowes VA, Wolfert RL,Bartsch G, Klocker H, Pointer J, Reissigl A, McLeod DG, Doug-las T, Morgan T, Kenny GM, Ragde H, Boynton AL, Holmes EH:Evaluation and comparison of two new prostate carcinomamarkers. Free-prostate specific antigen and prostate specificmembrane antigen. Cancer 1996;78:809–818.

11. Sanford E, Grzonka R, Heal A, Helal M, Persky L, Tyson I:Prostate cancer imaging with a new monoclonal antibody: Apreliminary report. Ann Surg Oncol 1994;1:400–404.

12. Carter RE, Feldman AR, Coyle JT: Prostate-specific membraneantigen is a hydrolase with substrate and pharmacologic char-acteristics of a neuropeptidase. Proc Natl Acad Sci USA 1996;93:749–753.

13. Troyer JK, Beckett ML, Wright GL Jr: Detection and character-ization of the prostate-specific membrane antigen (PSMA) intissue extracts and body fluids. Int J Cancer 1996;62:552–558.

14. Sobin LH, Wittekind CH (eds): ‘‘TNM Classification of Malig-nant Tumours,’’ 5th ed. New York: Wiley-Liss, 1997:170–173.

15. Douglas TH, Connelly RR, McLeod DG, Erickson SJ, Barren RJ,Murphy GP: The effects of exogenous testosterone replacementon prostate specific antigen and prostate specific membrane an-tigen levels in hypogonadal men. J Surg Oncol 1995;59:246–250.

16. Tillyer CR, Konings M, Gobin PT, Iqbal J: Disagreement be-tween the Roche Cobas Core and Hybritech Tandem-E PSAassays when measuring free, complexed and total serum pros-tate specific antigen. Ann Clin Biochem 1994;31:501–505.

17. Metz C: Basic principles of ROC analysis. Semin Nucl Med 1978;8:283–298.

18. Zweig M, Campbell G: Receiver-operating characteristic (ROC)plots: A fundamental evaluation tool in clinical medicine. ClinChem 1993;39:561–577.

19. Pfaffenberger RC, Paterson JH: ‘‘Statistical Methods for Busi-ness and Economics,’’ Richard Irwin, Inc., 1977.

20. Nelp WB: Anti-prostate antibody (ProstaScintt) imaging pre-and post-operatively. Eighth Annual Symposium, NorthwestHospital, June 13–14, 1997:169–170.

21. Chengazi VU, Feneley MR, Ellison D, Stalteri M, Granowski A,Granowski M, Nimmon CC, Mather SJ, Kirby RS, Britton KF:Imaging prostate cancer with technetium-99m-7E11.C5.3 (CYT-351). J Nucl Med 1997;38:675–682.

22. Gervasi LA, Muta J, Easley JD, Wilbanks JH, Seale-Hawkins C,Calton CE Jr, Scardino PT: Prognostic significance of lymphnodal metastases in prostate cancer. J Urol 1989;142:332–336.

23. Paulson DF: Impact of radical prostatectomy in the manage-ment of clinically localized disease. J Urol 1994;152:1826–1830.

24. Golimbu M, Morales P, Al-Askari S, Shulman Y: CAT scanningin staging of prostate cancer. Urology 1981;18:305–308.

25. Hricak H: Noninvasive imaging for staging of prostate cancer:Magnetic resonance imaging, computed tomography and ultra-sound. NCI Manager 1988;7:31–35.

TABLE V. Overall Correlations of ProstaScintT Scan With Serum PSMA or PSA Levels*

Preoperative

Stage Scan Postoperative scan

Serum RXa No RX Total RX No RX Total RX No RX Total

PSMA − + + − − − − + −PSA − − − − + + − − −N 48 38 86 110 40 150

*Range preoperative: PSMA, 0.049–0.567 ng/ml; PSA, 0.070–350.90 ng/ml. Range postoperative: PSMA, 0.130–0.774 ng/ml; PSA,0.000–204.30 ng/ml. Mean time postradical prostatectomy: RX, 2.40 years; No RX, 2.46 years; Total, 2.42 years.aRX, treatment.

Correlation of Serum PSMA and Prostate Scans 285