Niveaux-de-preuves-pour-lutilisation-du-PSA-et-des-autres-biomarqueurs-dans-la-dtection-prcoce-du-cancer-de-la-prostate-2014

Table 2: Characteristics of randomized clinical trials assessing PSA in the early detection of prostate cancerDetection / screening tests Serum processing and storage Definition of posit

Association Française d’Urologie Collège National de Biochimie des Hôpitaux Société Française de biologie Clinique Société française de médecine Nucléaire – Groupe de Biologie Spécialisée

GROUPE DE BIOLOGIE DE LA PROSTATE

Niveaux de preuves pour l’utilisation du PSA et des autres biomarqueurs dans la détection précoce du cancer de la

prostate

Review of evidence for the use of PSA and other

biomarkers in the early detection of prostate cancer

Coordonnateur du groupe de travail: Pierre-Jean Lamy1

pierre-jean.lamy@icm.unicancer.fr

Et

Anne-Sophie Gauchez2asgauchez@chu-grenoble.fr

Laurent Salomon 3laurent.salomon@hmn.aphp.fr

Margaret Haugh4 mhaugh@medicom-consult.com

CNBH

Agnès Georges7agnes.georges@chu-bordeaux.fr

Stéphane Larré8stephanelarre@yahoo.fr

Sylvain Loric9 sylvain.loric@hmn.ap-hop-paris.fr

Elisabeth Luporsi10e.luporsi@nancy.unicancer.fr

Pierre-Marie Martin11pierre-marie.martin@mail.ap-hm.fr

Catherine Mazerolles12 mazerolles.c@chu-toulouse.fr

Vincent Molinié13 vincent.molinie@chu-fortdefrance.fr

Pierre Mongiat-Artus14 pierre.mongiat-artus@sls.aphp.fr

Jacques Piffret15 jacques.piffret@orange.fr

François Thuillier16 thuillierfrancois@sfr.fr

Paul Perrin17paul.perrin@chu-lyon.fr

Xavier Rebillard18xavier.rebillard@wanadoo.fr , et le groupe de relecture du biologie

de la prostate : Drs, David Azria, Maguy Bernard, Karim Chick, Cyril Clavel,

Stéphane Culine, Olivier Cussenot, Alexandre de la Taille, Aurélien Descazeaud, David Guenet, François Iborra, Jacques Irani, Igor Latorzeff, Marie-Pierre Moineau, Didier Peiffert, Pierre Richaud, Jean Marc Riedinger, Pascal Rischmann, François Rozet, Corinne Sault, Virginie Vlaeminck-Guillem

1 Biologie Spécialisée et Oncogénétique, CRLC Val d’Aurelle, Montpellier France

2 UMR-S INSERM 1039, Institut de Biologie et de Pathologie CHU Grenoble, France

3 Service d'Urologie, APHP CHU Henri Mondor, Créteil, France

4 MediCom Consult, Villeurbanne, France

5 UMR_S 1113, FMTS, Université de Strasbourg, Strasbourg, France

6 Société Française de Médecine Nucléaire, Paris, France

7 Médecine Nucléaire, CHU Bordeaux, France

8 Service d'urologie, CHU de Reims, Reims, France

9 Biochimie clinique et génétique, APHP CHU Henri Mondor Créteil, France

10 Service d’Oncologie, Centre Alexis Vautrin, Nancy, France

Laboratoire de Transfert d’Oncologie Biologie, APHM, Marseille, France

12 Laboratoire d'Anatomie et Cytologie Pathologiques, IUCT Oncopole 1, Toulouse, France

13 Laboratoire d'Anatomie et Cytologie Pathologiques CHU La Meynard Fort de France, France

14 Service d'urologie, APH, CHU Saint-Louis, Paris, France

15 Association Française d’Urologie, Paris, France

16 Société Française de Biologie Clinique, Collège National de Biochimie des Hôpitaux Paris, France

17 Service d'urologie, Hôpital Lyon-Sud, Hospices Civils de Lyon, Lyon

18 Clinique Beausoleil, Montpellier, France

RESUME Contexte: Malgré des preuves contradictoires sur le bénéfice de l'utilisation del'antigène prostatique spécifique (PSA) pour la détection précoce du cancer de laprostate celui-ci est actuellement largement utilisé De nouveaux biomarqueursvisant à améliorer la valeur prédictive du PSA sont également utilisés.

Objectif: Examiner systématiquement les données scientifiques sur l'utilisation du

PSA et d'autres biomarqueurs pour la détection précoce du cancer de la prostate.Acquisition des données: nous avons cherché dans PubMed les essais cliniques etdes études publiés entre 2000 et mai 2013, évaluant le PSA et d'autresbiomarqueurs pour la détection précoce du cancer de la prostate, étude quicomprenait plus de 200 sujets Le niveau de preuve de l’utilité clinique a été évalué

en utilisant un système d’évaluation spécifique aux marqueurs tumoraux Un total de

84 publications, correspondant à 70 essais et études ont été sélectionnés pour cetterevue

Synthèse: Six essais cliniques randomisés évaluant le PSA ont été identifiés, maisquatre présentaient des faiblesses méthodologiques Bien que ces essais aientinclus un grand nombre de sujets avec un long suivi, leurs résultats présentent deslimites qui sont dues à la contamination (par des dosages antérieurs du PSA) dans legroupe de contrôle, à la qualité du suivi de ce groupe et de la variabilité dans lesméthodes Malgré ces limites, nous avons attribué un niveau de preuve IA (le plusélevé) pour l’utilisation du PSA pour la détection précoce, mais nous nerecommandons pas son utilisation dans le dépistage de masse Les biomarqueursémergents ont été évalués dans les études cas-témoins et de cohorte prospectives:PCA3 (n = 3); kallicréines (n = 3); [-2] ProPSA n = 5); oncogènes de fusion (n = 2).Ces études ont utilisé les résultats des biopsies pour le cancer de la prostate pour

déterminer la spécificité et la sensibilité des tests, mais elles n’ont pas évalué l'effetsur la mortalité Le niveau de preuve attribué était III-C, insuffisant pour uneutilisation en clinique.

Conclusions: Le PSA peut être utilisé pour la détection précoce du cancer de laprostate, mais le dépistage de masse n’est pas recommandé Les études surd'autres biomarqueurs suggèrent qu'ils pourraient être utilisés, individuellement ou encombinaison, pour améliorer la sélection des patients avec des niveaux élevés dePSA en vue de la réalisation d’une biopsie, mais des essais cliniques évaluant leurimpact sur la gestion du cancer de la prostate et sur la mortalité sont nécessaires

ABSTRACT

Context: Despite conflicting evidence for the benefit of using of prostate specificantigen (PSA) screening in the early detection of prostate cancer (PCa), is currentlywidely used New biomarkers aiming to improve the predictive value of PSA are alsoused

Objective: To systematically review the evidence for the use of PSA and otherbiomarkers in the early detection of prostate cancer

Evidence acquisition: We searched PubMed for clinical trials and studies assessingPSA and other biomarkers in the early detection of prostate cancer, publishedbetween 2000 and May 2013 that included >200 subjects The level of evidence(LOE) for clinical utility was evaluated using the tumor marker utility grading system

A total of 84 publications, corresponding to 70 trials and studies were selected for

Evidence synthesis: Six randomised controlled trials (RCTs) assessing PSA wereidentified but four were found to have methodological weaknesses Although thesetrials included large numbers of subjects and long-term follow-up, their limitationsinclude contamination in the control group, lower quality follow-up in this group andvariability in methods Despite these limitations, we attributed a level of evidence(LoE) of IA to PSA for early PCa detection, but we do not recommend its use in massscreening Emerging biomarkers were assessed in prospective case-control andcohort studies: PCA3 (n=3); kallikreins (n=3); [-2]proPSA n=5); fusion oncogenes(n=2) These studies used biopsy results for prostate cancer to determine specificityand sensitivity, but they did not assess the effect on PCa mortality The LoE attributedwas III-C.

Conclusions: PSA can be used for early prostate cancer detection but massscreening is not recommended Studies on other biomarkers suggest that they could

be used, individually or in combination, to improve the selection of patients withelevated PSA levels for biopsy, but RCTs assessing their impact on prostate cancermanagement and mortality are needed

Message pour les patients: Dans cette étude nous avons recherché quelles sontles preuves scientifiques pour l’utilisation du PSA et des autres biomarqueurs pourdétecter précocement un cancer de la prostate

Le PSA reste le marqueur diagnostique standard du cancer de la prostate Plus lePSA est élevé plus le risque du cancer est grand A des niveaux élevés (> 10ng/mL),

il y a un intérêt non discutable de rechercher un cancer par biopsie

Nous n’avons pas de preuves permettant de valider un dépistage de masse utilisant

le PSA Pour les autres biomarqueurs, les données suggèrent un intérêt de leurutilisation dans la sélection des patients avec un PSA élevé qui auraient besoin d’unebiopsie Néanmoins cela doit être confirmé dans des essais cliniques dédiés

Patient summary: In this study, we reviewed evidence for the use of PSA and otherbiomarkers for the early detection of prostate cancer

PSA remains the standard diagnostic biomarker in prostate cancer PSA level is acontinuous risk factor of prostate cancer At high levels (>10ng/mL) the need forfurther investigation, e.g biopsy is generally accepted

We did not find any evidence to support mass PSA screening Data for otherbiomarkers suggest that these may have a role to play in selecting subjects withelevated PSA levels for biopsy but well-designed RCTs assessing their impact onprostate cancer management are needed

Prostate cancer is the second most frequently diagnosed cancer in men worldwide(899 000 new cases, 13.6% of the total) and the fifth most common cancer overall.More than 70% of the cases occur in developed countries (644 000 cases) Theincidence of prostate cancer varies worldwide, with the highest rates observed inAustralia/New Zealand (104.2 per 100,000), and the lowest in South-Central Asia (4.1per 100 000) Incidence rates are relatively high in some developing regions such asthe Caribbean islands, South America and sub-Saharan Africa This variation can bepartly explained by differences in prostate specific antigen (PSA) testing andsubsequent biopsies that are more frequently performed in countries with the highestincidence

With an estimated 258 000 deaths in 2008, prostate cancer is the sixth cause ofcancer death in men (6.1% of the total) Because PSA testing has a much greatereffect on the incidence of prostate cancer than on its mortality rate, there is lessvariation in mortality rates worldwide (10-fold) than observed for incidence (25-fold).Prostate cancers, which occur mainly in older men, are often slow growing Theworldwide weighted mean of the age-specific rates (ASR (W)) for prostate cancermortality in developed regions is only twice that in developing regions (10.5 vs 5.6,respectively) These rates are generally higher in predominantly black populations(Caribbean, 26.3 per 100,000 and sub-Saharan Africa, 18-19 per 100 000), and verylow in Asia (e.g 2.5 per 100 000 in Eastern Asia) and intermediate in Europe, NorthAmerica and Oceania

In Europe the incidence of prostate cancer is 59.3 per 100 000, with a mortality rate

of 12.0 per 100 000 In France and the UK the mortality rates are similar, 12.7 and

13.8 per 100 000, respectively but the incidence is almost twice as high (118.3 and64.0 per 100 000, respectively) This difference in the apparent incidence of prostatecancer may be explained by the fact that in France 50% of men ≥50 years haveundergone PSA testing, compared with 10% in the UK.

The impact of PSA testing on the incidence of prostate cancer detection wasdemonstrated through the surveillance of the incidence of prostate cancer in the US[1] The introduction of PSA testing in the US resulted in an increase in the incidence

of prostate cancer from 140/100 000 in 1987 to 240/100 000 in 1993 After this peak,there was a reduction in the incidence of detected prostate cancer to 170/100 000,because there were no more ‘undetected’ cancers, just new cancers developing Thisincidence remains, nevertheless, higher than that prior to the introduction of PSAtesting in 1987

PSA is a glycoprotein that is produced mainly in the prostate epithelial cells Inhealthy men, PSA is generally concentrated in prostatic tissue and serum PSA levelsare very low PSA is an organ-specific marker rather than tumour specific and serumlevels can be elevated in the event of various non-malignant (e.g benign prostatichyperplasia (BPH), prostatitis, trauma) and malignant prostatic diseases Increasinglevels of serum PSA have been shown to be associated with a higher risk of prostatecancer [2] However, there is no clear-cut point that can differentiate insignificantcancers, which are not likely to be life-threatening, from those that are significant andlikely to be life-threatening [2]

Screening can help to detect chronic diseases and cancers in people before theydevelop symptoms with the aim of being able to offer treatment before it is too late In

listed criteria for screening Over the years, other criteria have been proposed butthey are mainly based on the WHO criteria [4] One of these recent criteria is thatthere should be strong scientific evidence that the screening program is effective From a public health point-of-view, the aim of prostate cancer screening is to improveoverall survival and from an individual point-of-view, the aim is to increase prostatecancer disease-free survival and decrease the prostate cancer-specific mortality rate.

In the USA prostate cancer screening began in the eighties when the majority ofpatients were diagnosed with advanced prostate cancer and a poor survival rate.Screening resulted in prostate cancer being detected at earlier stages and thereforethe patients had higher disease-free survival rates In addition, the lower PSAthreshold introduced and the increased number of biopsy cores resulted in thedetection and treatment of a large number of patients with low-volume and low-gradetumours [5] Prostate cancer screening is now widely used in the developed world,although the scientific evidence supporting its benefits is controversial One of thecontroversies concerns the choice between mass screening, i.e offering screening toall men after a certain age, or targeted or individual screening, i.e proposingscreening to men ‘at risk’, also known as ‘early detection' Another controversyconcerns the PSA cut-point since many studies have shown that there is a continuum

of prostate cancer risk at all values of PSA and no cut-point with a simultaneouslyhigh specificity and high sensitivity [6, 7] Even at low PSA levels, e.g 2.1 ng/mL, theresults from Prostate Cancer Prevention Trial (PCPT) showed a sensitivity of 52.6%and a specificity of 72.5% It is possible that some of the emerging biomarkers couldimprove the specificity of PSA and other prognostic factors and thus help in themanagement decision

As part of an overall review on the role of PSA and other emerging biomarkers indetection, diagnosis and treatment of prostate cancer, we present here a review ofthe literature on the use of PSA and other biomarkers in the early detection ofprostate cancer with an assessment of the level of evidence [8].

Methods

PubMed was searched using combinations of the terms given in Table 1 to identifypublications in English or French and published between 01/01/2000 and 06/05/2013that evaluated PSA and other biomarkers in prostate cancer This search was for aglobal systematic review to cover the use of biomarkers for early detection,prognosis, and prediction In this manuscript only the results for the use ofbiomarkers in the early detection of prostate cancer are presented The titles andabstracts of the references were screened to identify potentially pertinent referencesfor the global review A second screen on the titles and abstracts of referencesselected in the first screen was performed to identify potentially pertinent referencesfor the present analysis on early detection of prostate cancer Full papers were thenobtained and relevant papers were selected to correspond to the followingcategories:

Early detection with PSA in randomised clinical trials (RCTs)

Early detection with PSA in systematic reviews/meta-analyses

Early detection with PSA in ‘real-life’ studies

Early detection with PSA in prospective cohort studies

Early detection with emerging biomarkers

International guidelines for early detection of PSA

In addition, the co-authors were asked to provide details of studies they knew aboutthat had not been detected in the PubMed search and reference lists of included

randomized clinical trials or prospective cohort studies were included, since we knewthat large randomized clinical trials were available For the other biomarkers, weincluded randomized clinical trials and prospective and retrospective cohort studies.

We did not include studies assessing genetic testing since these require biopsytissue and are, therefore, not used in early detection of prostate cancer The studycharacteristics and results were extracted into tables and verified by at least two ofthe authors

Results

A total of 5825 publications were retrieved from the PubMed search and 75 additionalpublications were identified either by the authors or in the reference lists of includedstudies (Figure 1) A total of 2256 publications were selected after the initial screen

on titles and abstracts; these were then rescreened to select potential publications forthis review on biomarkers in early detection The PDFs were obtained for 254 articlesfor more detailed screening Finally, 70 studies were selected for inclusion in thisreview (Figure 1)

Total PSA in early detection of prostate cancer in RCTs

Although we identified five randomized clinical trials, only two were retained foranalysis because the other trials had major flaws in their design resulting in high risk

of bias In the Stockholm trial, the participants were not randomly selected and so arenot representative of the population [9, 10] In addition, this study was carried outmany years ago and used diagnostic methods that are no longer used Both theStockholm study and the Norrkoping study reported a 10-fold higher risk of prostatecancer than the other studies so that their results are not generalizable to othercountries [9-11] The results from the Norrkoping trial are not comparable with the

other trials because the criteria used was cytology [11] The study performed inQuebec was judged to have a risk of bias due to no concealment of allocation andresults not given from an intention-to-treat analysis [12].

The characteristics, quality assessment and results of the two trials included, theAmerican Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trialand the European Randomized Study of Screening for Prostate Cancer (ERSPC) aresummarized in Tables 2, 3 and 4 [13-40] The individual countries in the ERSPC trialhad their own specificities in terms of the methods used (Table 2) The PLCO trial didnot compare early detection with no early detection as there was a very high rate ofopportunistic screening in the ‘unscreened’ group This can be considered as acomparison of intense screening vs less intense screening [5, 41] In contrast, theERSPC trial, which was conducted in countries where PSA testing is generally farless prevalent, reported a reduction in prostate-cancer mortality Over time, the size

of the reduction has continued to increase; however, the number needed to screenand the number needed to detect to prevent one death from prostate cancer are1,055 and 37, respectively [14] An analysis of the data at 11 years suggest that forevery 1000 men who have a PSA test, 35.1 more cancers will be detected and 1additional prostate cancer death will be prevented It was calculated that to detectthese cancers in the screening group, 267 biopsies and 56 prostatectomies wouldhave to be performed

The 95% confidence interval for prostate cancer mortality for both trials overlap(0.87–1.36 for PLCO; 0.68–0.91 for ERSPC); the overlap of the 95% confidenceintervals, suggest that the trial results are not heterogeneous

Total PSA in early detection of prostate cancer: systematic review / analyses

meta-Two systematic reviews of RCTs assessing PSA in early detection have beenpublished (Table 5) [42, 43] The methods used in the trials included in these reviewsdiffered and many methodological concerns were reported No statistically significantreduction in prostate cancer-specific mortality was observed: RR=0.88 (95% CI=0.71– 1.09) and RR=0.95 (95% CI=0.85 – 1.07) There was significant heterogeneitywhen all trials were analysed together; this disappeared when the trials judged to be

at a high risk of bias were removed in one of the meta-analysis, but the reduction inprostate cancer-specific mortality remained non-significant: RR=0.89 (95%CI=0.77 –1.04) [43] The authors of one of the reviews concluded that there was no evidencefrom randomised clinical trials to support the routine use of screening for prostatecancer with PSA (with or without digital rectal examination (DRE) [42] Theconclusions on the other paper were that since any benefits from prostate cancerscreening may take >10 years to accrue, men with a life expectancy of <10-15 yearsshould be informed that screening for prostate cancer is not beneficial and can beharmful [43]

Total PSA in early detection of prostate cancer: ‘real-life’ studies

Three studies assessing PSA for early detection of prostate cancer in ‘real-life'settings were identified; one from the US, one from Austria and one from Japan(Tables 6 and 7) No benefit from PSA was observed in the American study, probablydue to the high contamination rate in the control group [44, 45] In contrast, in theAustrian study both the before/after comparison in the region where the screening

was performed and the comparison with the whole country showed a reduction inprostate cancer-specific mortality; 0.70 (95% CI=0.57-0.87) and 0.92 (95% CI=0.87-0.97), respectively [46, 47] In the Japanese study, 44% (1 224/2 775) of mendiagnosed with prostate cancer lived in communities with mass screening and theprostate cancer-specific mortality rates were 32.9% and 56.4% in the communitieswith and without mass screening, respectively [48].

Total PSA in early detection of prostate cancer: prospective cohort studies

We identified six prospective studies that assessed PSA for the early detection ofprostate cancer (Tables 8 and 9) [6, 49-55] These studies included between 300 and

39 213 participants (median: 12 215) and used different techniques for PSA testing(different tests and different thresholds) and different biopsy protocols Most of thesestudies confirmed that the risk of prostate cancer increased with increasing PSAlevels, although there was no consensus for the cut-point (Table 9) The authors ofone study suggested that there was a continuum of risk and that this should becommunicated to patients and healthcare professional [6, 50] The smallest studyreported interim results for 300 subjects with families with BRCA1/2 mutations andtherefore, a genetic predisposition The authors concluded that the results from theinterim analysis suggested that screening men with a genetic predisposition canallow clinically significant prostate cancer to be detected These results need to beconfirmed with the analysis of the whole cohort In June 2013, the study was stillrecruiting; 2620 subjects had been recruited [56]

The main recommendations from 13 international clinical practice guidelines,published from 2006 (one guideline) to 2012 (five guidelines) for PSA testing in theearly detection of prostate cancer are summarised in Table 10 [57-70] Theseguidelines were from the US (n=5), Canada (n=1), Japan (n=1), Europe (n=2), UK(n=2) France (n=1), and Belgium (n=1) None of the guidelines recommended massscreening, generally because the cost-benefit has not been demonstrated Themajority of the guidelines recommended individual testing for early detection insubjects aged <75 years old (with no comorbidities), with a life expectancy of morethan 10 years or in at-risk subjects (family history, ethnic origins, e.g Afro-WestIndians, Afro-Americans) They also stress the need for the subjects to be informedbefore PSA testing

The recommendation for early PSA testing in men aged from 40 to 50 years old isbecoming more widespread The rationale for this is that it can help to evaluate andstratify the prostate cancer risk and therefore adapt the frequency of future testingbased on the estimate risk of prostate cancer A PSA level of <1 ng/mL seems to begenerally accepted as a standard to define low risk of prostate cancer and thereforelower frequency or even stopping subsequent PSA testing

Summary of studies assessing PSA velocity (PSAV) and doubling time (PSADT)

A total of 1 RCT, 15 retrospective cohort studies and a systematic review wereidentified assessing the use of PSAV or PSADT for the early detection of prostatecancer were identified [50, 71-86] Data for patients in the RCT (PCPT – Finasteride

vs placebo) [50] suggested that PSAV was correlated with a higher risk of cancer atbiopsy in both the placebo and treatment groups In the placebo arm, the addition of

PSAV in multivariate analysis that included PSA, led to a very small increase in thearea under the curve of the receiving operating characteristic curve (AUROC) forpredicting positive biopsy (0.702 to 0.709).

Four of the retrospective cohort studies from the Rotterdam and Goteborg ERSPCstudy centers concluded that PSAV does not provide any advantage over PSA [74-77] The other 11 retrospective cohort studies concluded that PSAV was correlatedwith the risk of prostate cancer on biopsy and there was a correlation with the cancerstage [71-73, 78-85] Three of the studies also concluded that there was nocorrelation between PSADT and the risk of prostate cancer on biopsy [83-85]

One systematic review [86] analyzed 87 articles They concluded that here is littleevidence that PSAV and PSADT provide additional information above that provided

by total PSA (tPSA) alone

LOE for the use of PSAV for predicting positive biopsy: IIa (1 ancillary study of aRCT)

Harms of PSA-based screening for prostate cancer

Data about the harms of PSA-based screening for prostate cancer were reported bytwo of the randomized trials that assess the impact of PSA-based screening onprostate cancer mortality [14, 37] A false positive was defined as a positive resultand consequent workup with no histopathologic diagnosis of cancer within one year

of the screening test In the entire ERPSC trial, 75.9% of men that underwent abiopsy because of an elevated PSA value had a false-positive result [14]

In the PLCO trial, after four PSA tests, men had a 12.9% cumulative risk of receiving

at least one false-positive result (defined as a PSA level of ≥4.0 ng/mL and no

the potential psychological harms of prostate cancer screening, such as anxiety, or itsimpact on health-related quality of life.

Summary of studies assessing kallikrein 2 (hK2) in the early detection of prostate cancer

We did not identify any randomised clinical trials assessing the use of kallikrein 2(hK2) in the early detection of prostate cancer but we identified three prospectivecase-control studies (Table 11) [87-89] Two of these studies included men who hadbeen referred for biopsy, and the third included men referred because their PSA was

≥4 ng/mL or between, 3-4 ng/mL and they had an abnormal DRE In this latter studythe results from multivariate analyses showed that the risk of prostate cancerincreased with increasing concentrations of hK2 and hK2/fPSA [87] One of the otherstudies concluded that hK2 concentrations were statistically different in patients withbiopsy-confirmed prostate cancer while the other did not find a statistically significantdifference [88, 89] In this latter study the ratios f/tPSAS, hK2/fPSA and hK2/(f/tPSA),but not hK2/tPSA were all statistically significantly different in patients with prostatecancer compared with those with BPH [89] Thus, these studies, with the highestlevel of evidence identified, showed conflicting results for hK2 used alone Whenused in combination with other measures of PSA there seemed to be someadvantage, but the level of evidence remained low

In addition, we identified 10 retrospective cohort studies and 3 retrospective control studies (Table 10) The results from some of these studies suggest that hK2alone or in combination with other markers offers some advantages over PSA forselection men for biopsy However, there are conflicting results from other studies Inaddition, many of these studies assessed hK2 in populations with high rates of

case-prostate cancer, which makes extrapolation to a population undergoing primary earlydetection difficult.

LOE for kallikrein 2 (hK2) in the early detection of prostate cancer: conflicting datawith LOE<II

Studies assessing emerging biomarkers in early detection of prostate cancer

We identified 28 studies assessing other biomarkers: these studies were eitherprospective cohort or case-control studies (Table 12) [90-118] The biomarkersassessed most often were PCA3 (prostate cancer antigen 3) and PHI (ProstateHealth Index), a composite biomarker involving total PSA, free PSA and –(2)proPSA,

an isoform of free PSA) Generally, the populations included in the studies had highrates of prostate cancer, which makes extrapolation to a population undergoingprimary early detection difficult Many studies included men who had already had anegative biopsy as well as those undergoing first biopsy In addition, the biomarkerswere often used in combination with different forms of PSA and included in a logisticregression model or an artificial neural network The conclusions of the studies wereheterogeneous

Sixteen studies assessed the PCA3 score in prostate cancer diagnosis in eitherprospective cohort or case-control studies using a variety of methods to determinethe PCA3 concentration (Table 11) [101, 103, 104, 109, 111, 114, 116-118] Almost allstudies suggest that PCA3 performs better than PSA When compared with otheremerging biomarkers, the results were contradictory for PCA3 being better or not.The ranges for the positive predictive value (PPV) and the negative predictive value

studies do not provide evidence for the use of PCA3 for the early detection ofprostate cancer but the results suggest there may be role for PCA3 in deciding whichmen with elevated PSA levels should undergo biopsy.

LOE of PCA3 in patient selection for a second biopsy: II-b

Five studies assessed pro-PSA (Table 12) [99, 102, 107, 112, 113] These studies

concluded that pro-PSA, in combination with other forms of PSA, could improve thedetection of men more likely to have prostate cancer and thus reduce the number ofunnecessary biopsies However, the studies reported the AUROC and there were noclear cut-points The men included in these studies had a high level of PSA andtherefore had a high risk for prostate cancer; they were not representative ofpopulations that would be targeted for early detection

LOE for PHI in patient selection for a second biopsy: II-b

Five studies assessed the role of fusion proteins in men who had undergone PSA

testing and/or biopsy [104-106, 108, 115] One study reported a positive predictivevalue (PPV) of 0.94 and a negative predictive value (NPV) of 0.6 but 70% of thepatients had prostate cancer [104] Two studies reported that diagnostic accuracyincreased when the fusion proteins were combined with other biomarkers [105, 115]

In one of the other studies all the patients had prostate cancer and in the other thenegative controls were not necessarily biopsy-confirmed [106, 108] These results donot support the use of fusion proteins in the early detection of prostate cancer, butthey suggest that they could be useful in selecting men with elevated PSA levels forbiopsy The men included in these studies were pre-selected and therefore the fusionproteins were not used in the context of early detection screening

LOE for fusions genes in patient selection for a second biopsy: II-b

While the benefit of diagnosing cancer promptly in patients with symptoms with thehope of avoiding metastatic cancer and being able to provide effective treatment, isgenerally accepted, there is more controversy about detecting cancer inasymptomatic subjects Cancer detection in asymptomatic subjects can be doneeither through mass screen programs in which all subjects are invited to participate,

or through early detection programs in which subjects that may be at risk ofdeveloping a cancer are invited to participate The discovery that higher PSA levelswere associated with potentially lethal prostate cancer, which could be treated toincrease the likelihood of disease-free survival revolutionized prostate cancerprognosis Using lower thresholds results in improved prostate cancer diagnosis andthe detection of earlier-stage tumours with a higher chance of cure when treated.Following the initial euphoria of this breakthrough, concerns increased about theover-detection and over-treatment of an increasing number of indolent prostatetumours that were similar to those found during autopsy of men who died from otherdiseases [119] It has been estimated that the risk of having an indolent prostatecancer is approximately equal to the patient’s age minus 10, expressed as apercentage [120] Thus in Europe, where male life expectancy in 2010 was 72.5years, this risk is much higher than the lifetime risk of dying of prostate cancer, which

is estimated to be 3% [5, 121] In a long-term follow-up study of men with low-gradeprostate cancer 10-year prostate cancer survival was 97.2% supporting the low risk

of these cancers[122]

In a setting of increased PSA testing, several randomized clinical trials were initiated;

the first four years; in ERSPC which was a ‘federation’ of multiple screening trials indifferent countries, the screening frequencies differed between the trials and this wasnot always associated with DRE The results from PLCO did not show a reduction inprostate-cancer mortality although the rate of prostate cancer detection was higher inthe screening group [123] However, the control group was ‘contaminated’ by the highlevel of non-trial PSA testing in the community The level of this contamination waslower in the countries included in the ERSPC study The high level of pre-screeningwould have reduced the number of men with undetected prostate cancer that couldhave been detected in the trials, and would therefore have lowered the statisticalpower to detect a reduction in mortality [123].

The latest results from the ERSPC, after 11 years of follow-up, showed a lower rate

of prostate-cancer mortality in the screened group which continues to decrease overtime It was estimated that 1 055 men would have to be screened to prevent oneprostate-cancer death and to detect 37 prostate cancers [14] However, this wouldresult in 267 extra biopsies, 56 prostatectomies and high levels of adverse events(e.g incontinence, erectile dysfunction) For comparison, in two breast cancerscreening trials in Sweden and the UK it was estimated that 8.8 and 5.7 breast-cancer deaths, respectively would be avoided for every 1 000 women screened with4.3 and 2.3 per 1000 women over-diagnosed, respectively [124] This translates tobetween 2 and 2.5 lives saved for every over-diagnosed case In addition, it has beenestimated that 12% to 13% of screened men had false-positive results after three tofour screening rounds, and clinically important infections, bleeding, or urinaryretention occurred after 0.5%–1.0% of prostate biopsies [123]

As PSA is expressed by the normal prostate gland, it is not cancer specific biomarker.PSA levels increase with age as the prostate gland naturally becomes bigger; levels

are high in men with prostate cancer but they are also high in men with benignprostatic hyperplasia [125] PSA values are continuous but to be useful in clinicalpractice a threshold value needs to be defined When PSA levels are low, < 4ng/mLthe risk of prostate cancer is relatively low and when the levels are high, > 10ng/mLthe risk of prostate cancer is high and the clinical management of these patients isrelatively clear However, in the ‘grey zone’ between these levels there is uncertaintyabout the risk of prostate cancer The positive predictive value (PPV) for a PSAbetween 4.0 ng/mL and 10 ng/mL is about 25%, i.e less than one man in four willhave prostate cancer detected on biopsy, whereas that for a PSA > 10 ng/mL isbetween 42% to 64% (refs) In addition almost three out of four cancers detected inthe ‘grey zone’ are organ-confined with a good chance of being cured, whereas this

is less than 50% of those detected with PSA levels >10 ng/mL (ref) Hence, in thegrey zone, many unnecessary biopsies are performed because of the high falsepositive rate Although the threshold used often is 4 ng/mL, some laboratories havestarted to use a threshold between 2 and 3 ng/mL, without strong evidence [126,127] It is generally accepted that a single test for total PSA with a unique threshold isnot very informative and that it is better to assess its evolution over time, using thesubject as their own control This approach has been proposed for the early detection

of ovarian cancer Screening is based on patterns from serial CA-125 values topredict women at risk of preclinical ovarian cancer [128] Compared with the use ofsingle fixed cut-off CA-125 measurement, the model was shown to improvesensitivity for the early detection of ovarian cancer

Although the World Health Organisation (WHO) has developed international

obtained with the WHO international standard Even with use of the internationalstandard, quality control shows there is high intra- and inter-assay variability; thesecan be about 5% and up to 30%, respectively In addition to the assay variabilitythere are a number of factors that are known to affect the real level of PSA insamples, such as sport (cycling, jogging, horse-riding) sexual intercourse,manipulation of the prostate after biopsy, inflammatory infections of the urinary tract.Hence there are guidelines that should be respected to minimise the effect of thesefactors, including not to take samples within 48 hours of having practiced sport orhaving sexual intercourse and, above all, not to take samples from men withinflammatory conditions.

The main concerns of the high false positive rate are the cost and the harms thatarise from unnecessary biopsies Indolent prostate cancers that are either subject toactive surveillance or active treatment result in high costs, both medical and human.Active surveillance has economic implications for the healthcare system and thepatient and has human costs for the patient and their family with an impact on theirquality of life The high rate of indolent prostate cancers found that are treated leads

to significant side effects from the treatment and loss of quality of life for the patientand their partner [129]

The negative predictive value for PSA ≤4 ng/mL was estimated to be 85% in thePCPT [2] Among the men with PSA ≤0.5 ng/mL who underwent biopsy, 6.6%(32/449) had prostate cancer and 12.5% of these (4/32) had high grade cancer Therate of prostate cancer and high-grade cancers increased up to 26.9% (52/449) and25% (13/52), respectively in those with PSA between 3.1 and 4.0 ng/mL [2] Thesedata show that for all levels of PSA there is a risk of prostate cancer that increaseswith the PSA level

It seems reasonable to think that changes in PSA levels over time, (PSAV or PSADT)could be a better marker for more aggressive cancer compared with a singlemeasure of total PSA However, the results from the studies identified and the meta-analysis do not show they have any advantages over total PSA in the early detection

of prostate cancer

Although we know from the pre-PSA testing era, when prostate cancer was mainlydiagnosed by DRE, the lesions were frequently high-grade with high mortality rate.PSA testing has definitely enabled the detection of lower-grade, curable lesions Inthe PSA era, the incidence of prostate cancer has drastically increased with no suchincrease in prostate-cancer mortality Throughout the world, the incidence of prostatecancer varies widely It is higher in countries where PSA testing occurs, whileprostate-cancer mortality rates are more homogeneous

The challenge today is therefore to find a way to reduce the number of unnecessarybiopsies and reduce the detection of indolent prostate cancer, while ensuring thatpatients with lesions that require treatment are identified early In our review we foundmany studies that have attempted to increase the PPV by using other biomarkers forpatients with PSA levels in this ‘grey zone’ before deciding to perform biopsy.However, to date, there have been no randomised controlled trials assessing thesebiomarkers There have been many prospective cohort and case-control studiesperformed to assess the advantages of using hK2, fusion proteins, PCA3 and pro-PSA (Tables 11 and 12) Although the results were not always concordant, manyresults suggest that these emerging biomarkers could reduce the number ofunnecessary biopsies and improve the identification of men that should undergo

Some studies have evaluated the role of imaging for early detection of prostatecancer, but this was not the objective of this review Although prostate ultrasound hasnot been shown to be useful, results from prostate magnetic resonance imaging(RMI) assessment studies look promising for the detection or differentiation ofsignificant from insignificant prostate cancer [130].

In a setting where there is insufficient evidence to propose mass PSA screening,there does seem to be some merit to propose individual PSA testing [131] To identifymen who should be invited to undergo PSA testing, their risk factors should be takeninto consideration [132] In many guidelines, this approach to early prostate cancerdetection is suggested for men aged between 50 and 70 years with a life expectancy

of at least 10 years, with family history of prostate cancer, of a ‘high-risk’ ethnic andorganochlorine pesticide exposure (e.g chlordecone) (Table 10) [133] Although notwithin the scope of this review, it will be important to have a management strategy forthose men that are identified as being at risk of having prostate cancer in terms ofbiopsy or repeat PSA (and how regularly) Also outside the scope of this review butessential for the management of men who have positive biopsy results, we need tohave evidence about who should be offered active surveillance and who should beoffered active treatment

Conclusions

PSA has without doubt contributed to improve the prognostic of prostate cancer,enabling curable cancers to be identified, compared with the pre-PSA era While itremains the standard biomarker in prostate cancer, PSA level is a continuous riskfactor At high levels (> 10ng/mL) the need for further investigation, e.g biopsy isgenerally accepted However, when the levels are lower, even below the traditional

threshold of 4 ng/mL, other factors (presence of risk factors, results from DRE andpossibly from imaging, such as RMI) need to be taken into consideration in thedecision to propose biopsy with the aim of reducing over-diagnosis In the currentcontext where unorganized PSA testing in widely present and where variousbiomarkers are being developed for the early detection of prostate cancer, it wouldseem useful to suggest an individual management approach, integrating clinical andbiological data in order to obtain a more specific diagnosis and risk assessment Thisapproach could be used to choose between prostate biopsy and active surveillance.Mass screening is not to be recommended currently; it would seem that the futurelies with a personalised, multifactorial approach to early detection of prostate cancer

Acknowledgment: The authors are grateful to Julie Courraud for editing assistance This work was supported by an unconditional grant from Beckman-Coulter andHologic and by the scientific societies: Association Française d'Urologie, Collègenational de Biochimie des Hôpitaux, Société Française de Biologie Clinique, SociétéFrançaise de Médecine Nucléaire

Figure 1: Flow diagram of articles identified, screened and selected for inclusion in

current review A total of 70 trials and studies were included; these correspond to 84articles as some had more than one publication

Records identified through PubMed

searching (n = 5825)

Full-text articles excluded, with reasons (n = 174)

(n=4)

Full-text articles assessed for eligibility (n = 254)

Studies included in qualitative synthesis (n = 70)

Records to be screened for other topics (n = 2002)

Tables

Table 1: PubMed search: terms used in combination as either MeSh terms or free text

Prostatic Neoplasms / diagnosis / drug therapy / epidemiology / mortality /

radiotherapy / surgery / therapy

diagnosis / early diagnosis / prognosis

randomized controlled trial / controlled clinical trial / clinical trial / meta-analysis /

practice guideline / multivariate analysis / prognosis / prognostic / evidence-based

medicine*

PSA / pro-PSA / fusion oncogene / human PCA 3 / body mass index /

prostate-specific antigen / tumor burden

Table 2: Characteristics of randomized clinical trials assessing PSA in the early detection of prostate cancer

Detection / screening test(s)

Serum processing and storage

Definition of positive for further workup

Protocol for

‘positive’

Biopsy sample processing and storage and assessment

Outcomes measured

age: 55-69 years) in 8 European countries from

1993 to 2005

Serum PSA with Tandem-R/

Tandem-E Access assay (Hybritech)Screening interval = 4 years (Sweden =

2 years)External quality assessment(2 samples 6-

times/year)

Varied by center but in

2012 publication:

1995-1998:

≥3.4ng/mL1999-2004:

2.9ng/mLAfter 2004:

2.5ng/mL

Sextant prostate biopsy; lateralsextant biopsies adopted in June 1996

Central guidelines[134]

Placed in 10%

buffered formalin andsent to lab

neutral-in separately-numbered containersParaffin-blocked andH&E stainSemi-quantitative evaluation

of amount

of cancer

Primary: prostate cancer mortalitySecondary: overall mortality;

Detection / screening test(s)

Serum processing and storage

Definition of positive for further workup

Protocol for

‘positive’ Biopsy sample processing and

storage and assessment

Outcomes measured

N=9952 to screening and

109952 to control76.0%

screened

>once

All samples analysed in one

laboratoryProstatus;

Wallac Oy, Abo, FinlandScreening every two years; in third round, men with <1ng/mL found

in second round were not invitedMen ≥70 years no longer invited

After clotting, serum

obtained by centrifugation

at 3000 g for

20 mins and stored frozen

at -20°C within 3h fromsamplingPSA assay within 2 weeks of sampling and

<3 of thawing

≥3ng/mL, lowered to 2.54ng/mL from the third round

Medical history; DRE;

TRUS;

laterally directed sextant biopsyBlind review

of all medical records / pathology reports / autopsy protocols

Central guidelines[134]

Specimens step-sectioned in

4 mm incrementsTumor areas measured with 1mm grid; volumeestimated (multiplying the sum of tumor areas

in consecutivesections –

no correction for

shrinkage)

Absolute and relative reduction in cumulative prostate-cancer mortality Cumulative incidence of prostate cancerProportion of screening attendees

Detection / screening test(s)

Serum processing and storage

Definition of positive for further workup

Protocol for

‘positive’ Biopsy sample processing and

storage and assessment

Outcomes measured

74 years21,210 to screening;

21,166 to control94.6%

screened

>once

All samples analysed in one

laboratory (GP Lab) that coveredRotterdam and 7/12 of the

municipalitie

s (subjects from the remaining 5 were not included for PSA rates (only 11 tests)

In 1st round:

PSA-2 (Bayer); in 2nd round:

Hybritech Tandem E (Beckman-Coulter) (after 2000 replaced

Suspicious finding from any of the 3 diagnostic tests (PSA≥3ng/mL)

From November

1997 PSA threshold:

≥4ng/mL

laterally sextant needle biopsies

Central guidelines[134]

Placed in 10%

buffered formalin andsent to lab

neutral-in separately-numbered containersParaffin-blocked andH&E stainSemi-quantitative evaluation

of amount

of cancer in 4mm step sections

Primary outcome: prostate cancer mortality (blinded assessment

by Cause of Death Committee)

Detection / screening test(s)

Serum processing and storage

Definition of positive for further workup

Protocol for

‘positive’ Biopsy sample processing and

storage and assessment

Outcomes measured

2416; control:

1862) aged between 45-

70 years in MadridMailed invitation (population registry list)51.9%

screened

>once

All samples analysed in one

laboratoryScreening interval: 4 years

9 February 1996-12 May 1998: tPSA

>4ng/mL;

12 May

1998-1 January 2002: tPSA

>2.99ng/mL;

From 1 January 2002-present: tPSA

>2.99ng/mL and tPSA 1-2.99ng/mL and f/tPSA

≤20%

TRUS-guided sextant

prostate biopsy

Central guidelines[134]

Primary: prostate cancer mortality

Detection / screening test(s)

Serum processing and storage

Definition of positive for further workup

Protocol for

‘positive’ Biopsy sample processing and

storage and assessment

Outcomes measured

men aged

55-74 years (26.5% of invited participated)Randomisatio

n after informed consentScreening:

5188; control:

517190.7%

screened

>once

Initially all men had PSA, DRE and TRUS;

from 1996 only PSA and DREAll screening done at onecentre

1992-94:

suspicious finding from any of the 3 diagnostic tests (PSA≥10ng/mL)

1995-97:

same with threshold at

≥4ng/mL 1999-2003 PSA and DREwith

threshold:

≥3ng/mL

TRUS-guided prostate biopsy

Central guidelines[134]

Prostate cancer mortality city population database; prostate cancer (national and local cancer registries) blinded assessment

by Cause of Death Committee

Detection / screening test(s)

Serum processing and storage

Definition of positive for further workup

Protocol for

‘positive’ Biopsy sample processing and

storage and assessment

Outcomes measured

7286; control:

727149.9%

screened

>once

Screening interval: 4 years

currently:

DRE and TRUS guided biopsy

(transperineal)

Detection / screening test(s)

Serum processing and storage

Definition of positive for further workup

Protocol for

‘positive’ Biopsy sample processing and

storage and assessment

Outcomes measured

in South-west France (Tarn and Herault)

1035 in each group

Age 55-69 yearsDivided into previously screened/ no previous screen, then randomised within subgroup to screening or control50.1%

(Herault) and 49.0% (Tarn) screened

>once

Central testing (Beckmann-Hybritech)Every two years

PSA >3ng/mL DRE and

TRUS-guided sextant

biopsy; recall

of those with high-grade PIN, a negative biopsy with suspicious DRE/TRUS orPSA

>10ng/mL

Central guidelines[134]

Reduction of prostate cancer mortality

Detection / screening test(s)

Serum processing and storage

Definition of positive for further workup

Protocol for

‘positive’ Biopsy sample processing and

storage and assessment

Outcomes measured

metropolitan areas of Helsinki and TampereTotal 79,494 with 30,403 in screening groupMean age in both groups:

59.6 years74.4%

screened

>once

Central testing tPSA:

Hybritech Tandem-E (Beckman Coulter) andDelfia (Wallac) assays until June 2001 then Beckmann-Coulter assayf/tPSA:

Wallac ProStatus (Wallac)

Serum stored

at -80°C

4 aliquots stored and since 1998 whole blood samples also stored

PSA ≥4ng/mLPSA 3.0-3.9ng/mL : biopsy if suspicious DRE (1996-98) or if f/tPSA <0.16 (since 1999)

DRE, TRUS and TRUS-guided sextant biopsy(in 2002 – 10-12-core biopsy) from focal lesions, base, central and apex

Central guidelines[134]

Assessment done in fourparticipatinglaboratories

Prostate cancer mortality

Detection / screening test(s)

Serum processing and storage

Definition of positive for further workup

Protocol for

‘positive’ Biopsy sample processing and

storage and assessment

Outcomes measured

in Canton Aargau Randomisatio

n after informed consent

5150 in both groups50.0%

screened

>once

Screening interval: 4 yearsAxSym (Abbott) until June

2000 and Access (Beckmann-Coulter Hybritech from July 2000

Serum frozen

at -70°C within 2h of blood samplingSamples thawed immediately before PSA test within 2 weeks of sampling

PSA >3ng/mL

or PSA 3ng/mL and f/tPSA <20%

1-DRE and TRUS-guided transrectal sextant biopsytaken laterallyfrom the peripheral zone of the prostate

Central guidelines[134]

Each biopsy separately processed and

evaluatedAll positive and suspicious samples reviewed by

an independentreview pathologist

Detection / screening test(s)

Serum processing and storage

Definition of positive for further workup

Protocol for

‘positive’ Biopsy sample processing and

storage and assessment

Outcomes measured

38,343 screen;

38,350 control

in 10 centres across USA

Annual PSA for 6 years and DRE for

4 yearsSingle lab for testing using Tandem-R until Jan 1

2004 then Access Hybridtech PAS

Centralised laboratory

Serum frozen

at -70°C or colder within 2-4 hours of blood

collectionShipped weekly overnight on dry ice to central laboratorySpare samplestored at -70°C for future research

PSA:

≥4ng/mLDRE:

nodularity or induration, or investigator judgement of suspicion

No study- specific protocolLocal practice

in each study center

followedCertified tumor registrars ascertained stage, Gleason score and histology

Primary: prostate cancer mortalitySecondary: incidence; staging; survival

Table 3: PSA in randomized clinical trials: Trial design and quality characteristics

Study ID (ref) Randomisation

methods

Sample size calculation

Number of screened/invite

d (%)Number of lost

to follow-up

Median

follow-up time (min and max)

Blinding of outcome assessors

Contamination

of control group

Intention to screen analysis

ERSPC [13,

14, 16, 40]

Italy, France, Finland and Sweden:

randomisationbefore

informed consent usingpopulation registries –Belgium, Netherlands, Spain and Switzerland:

mailed invitation, then randomisationafter informedconsent

Power 80% todetect 25%

reduction in prostate cancer mortality with

10 years follow-up

Excluding France:

82.6%

underwent screening at least onceWith France:

63.4%

Median screen interval: 4.02 years

Mean and median follow

up (as of 2008) for coregroup

excluding France: 10.5 and 11.0 – including France: 8.6 and 9.8

available for Spain and Netherlands (others are on-going)Spain: in general population aged 55-69 years – 86.8 per 1000 patient-yearsNetherlands:

in population aged 55-69 years – 45 per

1000 years

patient-Yes on core age group (55-69 years old)