Jean-Philippe Charrier1

Carole Tournel1

Sandrine Michel1

Serge Comby2

Colette Jolivet-Reynaud3

Jacques Passagot1

Pascal Dalbon1

Denis Chautard4

Michel Jolivet1

1Département Rechercheet Développement,Unité Immunoessais

2Département Rechercheet Développement,Unité biomathématiques etbiostatistiques, BioMérieuxS.A., Marcy L’Etoile, France

3Unité mixte de recherche,103 bioMérieux/CentreNational de la RechercheScientifique, ENS,Lyon, France

4Service d’Urologie,Centre HospitalierUniversitaire d’Angers,Angers, France

Differential diagnosis of prostate cancer andbenign prostate hyperplasia using two-dimensionalelectrophoresis

Prostate specific antigen (PSA) is a protease which is characteristic of the prostate. It iswidely used as a serum marker for the early diagnosis of prostate cancer (PCa). Never-theless, for concentrations between 4 and 10 ng/mL, PSA does not enable PCa to bedistinguished from benign diseases, such as benign prostate hyperplasia (BPH). Insera, the use of a ratio between free PSA (PSA uncomplexed with protease inhibitor)and total PSA (free PSA and PSA bound to alpha-1 anti-chymotrypsin) enables the“gray zone” to be reduced, but an important proportion of patients are still wronglyclassed. Using two-dimensional electrophoresis, we demonstrated using 52 PCa and40 BPH well-documented clinical cases that BPH sera show a significantly greaterpercentage of low-molecular-weight free PSA elements (lwPSA) than PCa sera. In ourstudy, the use of a ratio between lwPSA and standard free PSA enables the correctdiagnosis of 100% of PCa and 82.5% of BPH cases as against when 73.1% and42.5% respectively were correctly diagnozed using the total PSA and the free/totalPSA ratio. This important finding may be related to differences in the mechanismsecreting PSA from the prostate into the bloodstream. We have shown how a tissuemarker may be turned into a powerful tumor marker by events probably unrelated to itsexpression.

Keywords: Two-dimensional polyacrylamide gel electrophoresis / Prostate-specific antigen /Prostate cancer / Benign prostate hyperplasia / Diagnosis EL 4408

1 Introduction

Prostate cancer (PCa) is a major public health problem. Inmen and in Western countries, it is the most commonlydetected cancer and the second leading cause of deathdue to cancer [1]. Since the 80’s, the use of prostate-spe-cific antigen (PSA) has improved the number of patientsdiagnozed at an early clinical stage [2]. This screeningpractice is showing its first benefits [1, 3] but suffers fromseveral drawbacks. Using a traditional cutoff of 4 ng/mL,38–48% with organ-confined but clinically significant PCahave a normal value [4, 5]. Moreover, 65% of patients withan elevated PSA concentration are affected by a nonma-lignant disease such as benign prostate hyperplasia(BPH) [6, 7]. These persons may potentially be candidatesfor a confirmation by biopsy, with the corresponding con-sequences in terms of health costs and possible mor-bidity.

The quest to find a marker able to differentiate betweenPCa and BPH is therefore of huge importance. To achievethis aim, many works have been made in PSA biochemis-try. It has notably been shown that PSA is a kallikrein ableto bind massively to protease inhibitors such as �-1 anti-chymotrypsin (ACT) [8, 9]. Advantage was taken of thisproperty to improve the differentiation between PCa andBPH. In serum, a ratio between free PSA (PSA uncom-plexed with protease inhibitor) and total PSA (free PSAand PSA bound to ACT) has been shown to reducethe number of unnecessary biopsies by 20% in the4–10 ng/mL total PSA zone [6].

This phenomenon is still largely unexplained but its eluci-dation is promising new improvements in the differentia-tion between PCa and BPH. To pursue this goal, we haverecently shown that it is possible to directly observe PSAin the serum of PCa or BPH patients using two-dimen-sional electrophoresis (2-DE) [10]. This technique allowsPSA detection up to a concentration of 100 pg/mL anddisplays both PSA bound to ACTand free PSA in a singleexperiment. Free PSA was shown to be formed by spotswith the standard 32 kDa molecular mass (swPSA) and byspots with lower molecular mass (lwPSA). Our previousresults suggested a higher proportion of lwPSA in BPHcompared with PCa cases. In this paper, we confirmthese findings in a study using 52 PCa and 40 BPH well-documented clinical cases.

Correspondence: Jean-Philippe Charrier, bioMérieux,69280 Marcy L’Etoile , FranceE-mail: jean-philippe_charrier@eu.biomerieux.comFax: +33-4-7887-2101

Abbreviations: BPH, benign prostate hyperplasia; fPDA, freeprostate-specific antigen; lwPSA, low molecular weight freePSA; PCa, prostate cancer; PSA, prostate-specific antigen;swPSA, standard molecular weight free PSA; tPSA, total pros-tate-specific antigen; ROC, receiver-operating characteristic

Electrophoresis 2001, 22, 1861–1866 1861

WILEY-VCH Verlag GmbH, 69451 Weinheim, 2001 0173-0835/01/0905–1861 $17.50+.50/0

Proteomicsand2-DE

2 Materials and methods

2.1 Epitope mapping

Overlapping peptides were simultaneously synthesizedon a nitrocellulose membrane utilizing Fmoc amino acidchemistry (SPOTs Synthesis, Cambridge Research Bio-chemicals, Cambridge, UK) according to Frank and Dör-ing [11]. Peptides were generated each in nmol quantitiessuitable for immunological detection. Antibody reactivityto the membrane-bound peptides was then analyzed byan indirect colorimetric immunoassay as previouslydescribed [12]. Spots corresponding to peptides withantibody reactivity produced a positive blue signal.

2.2 Sample collection

Serum samples were kindly provided by biologists andurologists in agreement with French legislation. The 52PCa patients have a mean age of 70.6 years (range 43–93 years) and the 40 BPH patients have a mean age of69.2 years (range 48–92 years). The diagnoses weremade by the referring physicians according to standardclinical and biochemical criteria. For this purpose, totalPSA (tPSA) and free PSA (fPSA) were determined using,respectively, the Vidas TPSA immunoassay kit and theVidas FPSA immunoassay kit from bioMérieux (MarcyL’Etoile, France). Depending on the availability of theinformation, the gleason grade or the stage of the tumorare shown in Table 1. Serum samples were kept frozenuntil analysis.

Table 1. Gleason grade or stage of the tumor for PCapatients

Gleason grade Number of patients

5 36 77 178 39 3

Stage of the tumor Number of patients

T1 0T2 10T3 4T4 5

2.3 2-DE and immunodetection of PSA

Sera were prepared and run as previously described indetail [10]. Briefly, 100 �L of serum were diluted with therehydratation solution (8.3 M urea, 2 M Thiourea, 4%

CHAPS, 100 mM DTT, 4–9 Servalyt 2%, 1 mg/L orangeG) and loaded into a nonlinear IPG strip pH 3–10 (Pharma-cia Biotech, Uppsala, Sweden). The samples were sepa-rated by 2-DE according to Bjellqvist et al. [13] with somemodifications [10]. The 2-D gels were transferred ontoPVDF membranes (Millipore, Bedford, MA, USA). Themembranes were incubated with the 13C9E9D6G8anti-PSA monoclonal antibody (bioMerieux, France) andwith a peroxidase-labeled anti-mouse antibody (JacksonImmuno-Research Laboratories, West Grove, PA, USA).Except for the last wash, all immunodetection stepswere performed in the presence of blocking solution(15 mM Tris, 0.14 M NaCl, 0.5 ml/L Tween 20, pH 8.0, 5%nonfat milk). PSA was then detected by chemilumines-cent using the Fluor-S system (Bio-Rad, Laboratories,Hercules, CA, USA) and Super Signal West DuraExtended Substrate (Pierce Chemical, Rockford, IL,USA). Both swPSA and lwPSA spots were later selected,quantified, and subtracted from the local backgroundusing the Multi-Analyst software (Bio-Rad).

2.4 Data analysis

Receiver-operating characteristic (ROC) curves wereplotted to assess the sensitivity and specificity of tPSA,the fPSA/tPSa ratio and the lwPSA/swPSA ratio. Thesethree values were compared using the areas under theROC curves method [14] for patients in the 4–10 ng/mLtPSA zone and for all cases. A p-value of less than 0.05was considered statistically significant in any test.

3 Results and discussion

3.1 Characterization of the anti-PSA13C9E9D6G8 antibody epitope

Overlapping nonapeptides offset by three covering thePSA sequence showed that the mAb 13C9E9D6G8reacted strongly with the peptides IEPEEFLTP (136–144)and EEFLTPKKL (139–147). Therefore, the deduced mini-mal sequence recognized by the mAb 13C9E9D6G8 wasEEFLTP (139–144). These results were then confirmedwith overlapping nonapeptides offset by one. This linearepitope was located just before the major PSA cleavagesite detected in sera (K145/K146) [15, 16]. The epitopesequence is not affected by subsequent cleavages [17]and may even be more accessible on PSA cleaved atresidues K145/K146. As a result, 13C9E9D6G8 antibodyhas shown an exceptional ability to detect cleaved PSAusing 2-DE [10]. As regards to the different cleavage sitesof PSA, 13C9E9D6G8 should essentially recognize theN-terminus end of the fPSA fragments [10].

1862 J.-P. Charrier et al. Electrophoresis 2001, 22, 1861–1866

Figure 1. fPSA 2-DE pattern ofPCa and BPH serum. Charac-teristic pattern of (A) PCa and(B) BPH. (C)–(F) rare pattern.

3.2 Images of fPSA in sera

By 2-DE, we were able to analyze PSA spots in the seraof 52 patients suffering from PCa and 40 patientsaffected by BPH. Figures 1A and B show the character-istic patterns of fPSA found in cases of PCa and BPH.These images match the large majority of cases understudy. Generally only one lwPSA spot was observed,and less often two or more (Figs. 1C and D). Other pro-files have also been observed previously [10] and cur-rently but in lower proportions. Figure 1E shows notably

the presence of polypeptides with a molecular masssuperior to the swPSA (� 36 000 Da). This pattern wasalso found in a small number of seminal plasma fromhealthy donors (data not shown), and was observed, inthe present study, in four PCa sera. Figure 1F shows theexistence of spots which are slightly heavier and morebasic than swPSA. These spots may correspond to thepro-PSA; they are found in three PCa cases and fourBPH serum. All spots identified as swPSA or lwPSAaccording to the patterns of Fig. 1 were quantified toestablish the lwPSA/swPSA ratio.

Electrophoresis 2001, 22, 1861–1866 Differential diagnosis of PCa and BPH using 2-DE 1863

Figure 2. Differentiation of PCa and BPH cases through(A) fPSA/tPSA ratio and (B) lwPSA/swPSA ratio.

3.3 Differentiation between PCa and BPH cases

The large number of patients involved in this study rein-forces our previous observations and the recent work ofHilz at al. [16]. The ratio lwPSA/swPSA is clearly more dis-criminating than the usual ratio fPSA/tPSA (Fig. 2). This istrue for patients in the 4–10 ng/mL tPSA zone and allpatients, although the fPSA/tPSA ratio is essentially ofuse in the 4–10 ng/mL tPSA zone. Thus, using a cutoff of6.5%, we were able to correctly diagnoze 100% of PCaand 82.5% of BPH cases compared with 73.1% and42.5%, respectively correctly diagnozed using the tPSAand fPSA/tPSA approach.

We preferred to choose a lwPSA/swPSA cutoff enablinga sensitivity as high as possible for PCa detection. Inour opinion, it is of huge importance to correctly diag-noze each significant PCa at an early stage. In this way,the risk of missing potentially curable cancer is over-come for the well-being of the patients within their lifespan. Even with this cutoff choice, the lwPSA/swPSAratio enables a 2-fold increase in the number of cor-

rectly classified BPH cases. This technique thereforebears the promise to avoid around 50% of unnecessarybiopsies.

As regards the seven misclassified BPH cases (17.5%), itshould be considered that roughly 25% of benign biop-sies have been found to be incorrectly negative [18]. Wemay therefore suppose that a small number of personsdiagnozed as BPH are in fact affected by PCa. Thatmeans even up to 50% of unnecessary biopsies may behoped to be avoided. Unfortunately, it was not possible toconfirm this assumption by recalling each discrepant BPHpatient. Nevertheless, special attention will be paid tothem during their next regular visit to their usual physi-cians. The preceding findings were confirmed by theROC curves. They undoubtedly show a higher diagnosticvalue for the lwPSA/swPSA ratio (Fig. 3). The differencebetween the areas under the curves (Table 2) demon-strate lwPSA/swPSA to be statistically more discriminat-ing than fPSA/tPSA for all the data considered (p�0.001)and to be virtually more significant in the 4–10 ng/mLtPSA zone (p = 0.0580).

Table 2. Areas under the ROC curves

All patients Patients with4 � tPSA � 10 ng/mL

Area underthe ROCcurves

Standarderror

Areas underthe ROCcurves

Standarderror

tPSA 0.884 0.0340 0.691 0.0749fPSA/tPSA 0.773 0.0495 0.757 0.0686lwPSA/swPSA 0.931 0.0303 0.906 0.0415

3.4 Nature of lwPSA

The nature of lwPSA is largely unknown. Currently, twomain hypotheses are emerging: lwPSA could be frag-ments resulting from the internal cleavage of PSA orPSA-related proteins expressed from alternative tran-script of PSA-DNA. The very low amount of lwPSA inserum makes the direct investigation of these hypothesesdifficult, include with the latest improvements of massspectrometry techniques. An alternative has been topurify PSA by immunoaffinity from a pool of sera orfrom blood aspirated during transvesical prostatectomy[15, 16]. This has demonstrated the existence of PSAcleaved at the position K145/K146 in both PCa and BPHcases. The 1–145 fragment has a theoretical pI of 6.04and a theoretical molecular mass of 16 kDa plus the N-glycosylation weight, consequently, it could correspondto the lower lwPSA spot of Figs. 3A and B. The other clea-vage site previously described on PSA from seminal liquid[17] may also lead to additional lwPSA spots.

1864 J.-P. Charrier et al. Electrophoresis 2001, 22, 1861–1866

Figure 3. ROC curves of tPSA, fPSA/tPSa ratio andlwPSA/swPSA ratio for patients (A) in the 4–10 ng/mLtPSA zone and (B) for all the cases.

On the other hand, PSA transcripts produced by alterna-tive splicing have recently been described [19, 20]. One ofthese transcripts has been cloned and expressed ineukaryotic cells [19], while the other transcript has beendetected in both BPH and PCa tissues [20]. It still remainsto be demonstrated if the related proteins are reallyexpressed in vivo, but they include the epitope recog-nized by our antibody and have a pI and a molecularmass possibly compatible with lwPSA spots. Finally, thehypotheses of a partial deglycosylation or an abnormalglycosylation should not be ruled out. The pattern ofswPSA with 3 to 5 spots, possibly related to glycosylationheterogeneity [21], is generally not observed on lwPSAspots. It is therefore possible to suppose the glycosyla-tion heterogeneity has been lost for most of the lwPSAelements.

3.5 Why is lwPSA significant in differentiatingbetween PCa and BPH?

The reason why lwPSA is relevant in distinguishing PCafrom BPH cases is totally unknown. Curiously, lwPSAseems to be much more abundant in prostate tissueand seminal fluid than in serum [10, 22]. This observationshould be related to the lack of correlation between thequantity of PSA present in the prostate and in the serum[22]. An explanation for these discrepancies will requirefurther investigations, but it may be possible to antici-pate on the importance of the secretory pathways. PSAis normally secreted into the glandular ducts through aunique apocrine mechanism. Prostatic secretory gran-ules, originating from the Golgi, accumulate in the apicalthird of the cell and are released in packets into the glandlumen [23]. In BPH tissue, PSA seems to diffuse “back-wards” from the lumen into the extracellular compart-ment and the circulation through epithelium disruption[24]. On the contrary, in PCa the polarization of the cellsis lost and the entire secretory apparatus is generallyabsent. PSA is released free into the cytoplasm andtend to be more sharply localized along the surface cellmembrane [23]. This phenomenon may lead to a leakageof the PSA directly into the extracellular compartment,followed by diffusion into the circulation. This differentleakage mechanism may therefore be the key point inunderstanding the difference in the lwPSA proportion inserum. Moreover, these observations support thehypothesis of a higher cleavage of PSA during entrap-ment in cystic-like dilated BPH glands for a long periodof time [25], when PSA, from PCa cells may rapidly reachthe bloodstream and as a result be less affected by pro-tease action.

4 Concluding remarks

We have demonstrated that the ratio lwPSA/swPSAenables excellent discrimination between PCa and BPHpathologies. Interestingly, PSA is not in the strict senseof the term a cancer marker, but a tissue marker. Indeed,PSA is almost exclusively produced by the epithelial cellsof the prostate whether the tissue is normal or malignant,and its presence in serum does not indicate a cancer butan alteration of the prostate tissue. Moreover, the con-centrations of PSA in serum and in prostate tissue arenot correlated. PSA is expressed in greater quantities inBPH tissue than in PCa tissue, whereas PSA may reachhigher concentrations in PCa than in BPH serum. Themarker property of PSA therefore seems to be related tothe modification of the gland architecture, rather than anabnormal expression into the tumor cells. Similarly, weanticipate that the difference in the pathway mechanisms

Electrophoresis 2001, 22, 1861–1866 Differential diagnosis of PCa and BPH using 2-DE 1865

described between PCa and BPH tissue may result in ahigher proportion of lwPSA in BPH patients than in PCapatients.

We have therefore shown how a tissue marker such asPSA may be turned into a powerful tumor marker byevents probably unrelated to its expression. The ratiolwPSA/swPSA will require the development of specificimmunoassays, but will drastically improve the detectionof early curable PCa while avoiding unnecessary prostatebiopsies. As a consequence, we hope that patients willtake advantage of easier management of their prostatediseases at lower health costs.

We wish to thank Drs. A. Daver, J. Guillet, G. Ricolleau,N. Piga, and T. A. Stamey for providinguswith human sam-ples, Drs. N. Battail-Poirot and G. Sibaï’s teams for provid-ing us with the purified monoclonal antibody against PSAand C. Dupuy and C. Micolaud for assistance with thetranslation. We are especially grateful to C. Courte for hisadvice and support during this study.

Received September 4, 2000

5 References

[1] Greenlee, R. T., Murray, T., Bolden, S., Wingo, P. A., CA Can-cer J. Clin. 2000, 50, 7–33.

[2] Stamey, T. A., Yang, N., Hay, A. R., McNeal, J. E., Freiha, F. S.,Redwine, E., N. Engl. J. Med. 1987, 317, 909–916.

[3] Merrill, R. M., Stephenson, R. A., J. Uro. 2000, 163, 503–510.[4] Lange, P. H., Ercole, C. J., Lightner, D. J., Fraley, E. E., Ves-

sella, R., J Urol. 1989, 141, 873–879.[5] Hudson, M. A., Bahnson, R. R., Catalona, W. J., J Urol. 1989,

142, 1011–1017.[6] Catalona, W. J., Partin, A. W., Slawin, K. M., Brawer, M. K.,

Flanigan, R. C., Patel, A., Richie, J. P., deKernion, J. B.,Walsh, P. C., Scardino, P. T., Lange, P. H., Subong, E. N., Par-son, R. E., Gasior, G. H., Loveland, K. G., Southwick, P. C.,JAMA 1998, 279, 1542–1547.

[7] Catalona, W. J., Smith, D. S., Ratliff, T. L., Basler, J. W.,JAMA 1993, 270, 948–954.

[8] Lilja, H., Christensson, A., Dahlén, U., Matikainen, M.-T.,Nilsson, O., Pettersson, K., Lövgren, T., Clin. Chem. 1991,37, 1618–1625.

[9] Stenman, U.-H., Leinonen, J., Alfthan, H., Rannikko, S., Tuh-kanen, K., Alfthan, O., Cancer Res. 1991, 51, 222–226.

[10] Charrier, J. P., Tournel, C., Michel, S., Dalbon, P., Jolivet, M.,Electrophoresis 1999, 20, 1075–1081.

[11] Frank, R., Döring, R., Tetrahedron 1988, 44, 6031–6040.[12] Jolivet-Reynaud, C., Dalbon, P., Viola, F., Yvon, S., Para-

nhos-Baccala, G., Piga, N., Bridon, L., Trabaud, M. A., Bat-tail, N., Sibaï, G., Jolivet, M., J. Med. Virol. 1998, 56, 300–309.

[13] Bjellqvist, B., Pasquali, C., Ravier, F., Sanchez, J.-C., Hoch-strasser, D., Electrophoresis 1993, 14,1357–1365.

[14] Hanley, J. A., McNeil, B. J., Radiology 1983, 148, 839–843.[15] Noldus, J., Chen, Z., Stamey, T. A, J. Urol. 1997, 158, 1606–

1609.[16] Hilz, H., Noldus, J., Hammerer, P., Buck, F., Luck, M.,

Huland, H., Eur. Urol. 1999, 36, 286–292.[17] Watt, K. W., Lee, P.-J., M’Timkulu, T., Chan, W.-P., Loor, R.,

Proc. Natl. Acad. Sci. USA 1986, 83, 3166–3170.[18] Stamey, T. A., Yemoto, C. E., J. Urol. 2000, 163, 119–126[19] Heuze, N., Olayat, S., Gutman, N., Zani, M. L., Courty, Y.,

Cancer Res. 1999, 59, 2820–2824.[20] Tanaka, T., Isono, T., Yoshiki, T., Yuasa, T., Okada, Y., Cancer

Res. 2000, 60, 56–59.[21] Prakash, S., Robbins, P. W., Glycobiology 2000, 10, 173–

176.[22] Jung, K., Brux, B., Lein, M., Rudolph, B., Kristiansen, G.,

Hauptmann, S., Schnorr, D., Loening, S. A., Sinha, P., Clin.Chem. 2000, 46, 47–54.

[23] Cohen, R. J., McNeal, J. E., Edgar, S. G., Robertson, T.,Dawkins, H. J., Hum. Pathol. 1998, 29, 1488–1494.

[24] Anim, J. T., Udo, C., John, B., Acta Histochem. 1998, 100,439–449.

[25] Chen, Z., Chen, H., Stamey, T. A., J. Urol. 1997, 157, 2166–2170.

1866 J.-P. Charrier et al. Electrophoresis 2001, 22, 1861–1866