The Long Journey of Cancer Biomarkers from the Bench to [617923]

The Long Journey of Cancer Biomarkers from the Bench to
the Clinic
Maria P. Pavlou,1,2Eleftherios P. Diamandis,1,2,3and Ivan M. Blasutig2,3*
BACKGROUND :Protein cancer biomarkers serve multi-
ple clinical purposes, both early and late, during diseaseprogression. The search for new and better biomarkershas become an integral component of contemporarycancer research. However, the number of new bio-markers cleared by the US Food and Drug Administra-tion has declined substantially over the last 10 years,raising concerns regarding the efficiency of thebiomarker-development pipeline.
CONTENT :We describe different clinical uses of cancer
biomarkers and their performance requirements. Wealso present examples of protein cancer biomarkerscurrently in clinical use and their limitations. The ma-jor barriers that candidate biomarkers need to over-come to reach the clinic are addressed. Finally, the longand arduous journey of a protein cancer biomarkerfrom the bench to the clinic is outlined with anexample.
SUMMARY :The journey of a protein biomarker from the
bench to the clinic is long and challenging. Every stepneeds to be meticulously planned and executed to suc-ceed. The history of clinically useful biomarkers sug-gests that at least a decade is required for the transitionof a marker from the bench to the bedside. Therefore, itmay be too early to expect that the new technologicaladvances will catalyze the anticipated biomarker revo-lution any time soon.
© 2012 American Association for Clinical Chemistry
The search for cancer biomarkers has become an inte-gral component of cancer research because of the po-tential of biomarkers to enable early detection of dis-eases and to provide diagnostic, prognostic, andpredictive information. According to the NIH, a bio-marker is defined as “a characteristic used to measure
and evaluate objectively normal biological processes,pathogenic processes, or pharmacological responses toa therapeutic intervention” (1). During the last decade,
the maturation of high-throughput -omics technolo-gies (i.e., genomics, transcriptomics, proteomics, pep-tidomics, and metabolomics) has set the pace for bio-marker discovery to a point whereby it has evolvedfrom being an observational byproduct of clinical prac-tice, toward being a large-scale systematic process, of-ten referred to as a “pipeline.” Interestingly, despite theintensified interest and investment by major stake-holders, including academia, industry, and govern-ment, the number of biomarkers receiving US Foodand Drug Administration (FDA)
4clearance has de-
clined substantially over the last 10 years to less thanone protein biomarker per year (2, 3) .
A simplistic version of a biomarker development
pipeline could be divided into 4 main phases (Fig. 1).First, preclinical exploratory studies are performed toidentify promising biomarkers. Usually during this dis-covery phase, small numbers of samples from diseasedand nondiseased groups are compared to identifymolecules exhibiting discriminating potential. Whenproperly performed, high-throughput technologiesenable the simultaneous unbiased assessment of thou-sands of molecules from which potential biomarkerscan be identified. The candidate biomarkers are prior-itized on the basis of various selection criteria that de-pend on the discovery platform, the type of biomarkerbeing sought, and the available bioinformatics tools.Judging by the numerous publications reporting novelcandidate biomarkers, the discovery phase seems to beproductive, and numerous reviews have summarized ageneralized strategy regarding this first phase of thepipeline (4–7) .
The second step includes the development and
validation of a robust assay to measure the analyte ofinterest in the intended clinical sample. In practice, as-say development is an iterative process that occurs atevery step in the pipeline and may not end even after an
1Department of Pathology and Laboratory Medicine, Mount Sinai Hospital,
Toronto, ON, Canada;2Department of Laboratory Medicine and Pathobiology,
University of Toronto, Toronto, ON, Canada;3Department of Clinical Biochem-
istry, University Health Network, Toronto, ON, Canada.
* Address correspondence to this author at: Toronto General Hospital, 200
Elizabeth Street, Rm. 3 EB 362, Toronto, ON, M5G 2C4. Fax 416-340-4215;e-mail ivan.blasutig@uhn.on.ca.
Received March 6, 2012; accepted June 28, 2012.Previously published online at DOI: 10.1373/clinchem.2012.1846144Nonstandard abbreviations: FDA, US Food and Drug Administration; TNM,tumor-node-metastasis; PSA, prostate-specific antigen; ER, estrogen receptor;CA19-9, carbohydrate antigen 19-9; uPA, urokinase-type plasminogen activa-tor; PAI-1, plasminogen activator inhibitor-1.Clinical Chemistry 59:1
147–157 (2013) Reviews
147

assay is marketed. The assay is used to quantify the
biomarker, usually in a retrospective fashion, using pa-tient samples stored in tumor banks, which is the thirdstep of the pipeline. In contrast to the discovery phase,retrospective validation studies require large numbersof samples to ensure statistical rigor, as well as samplesthat reflect the biological variability of the targetedpopulation. Only a very limited number of analytesthat continue to show discriminatory potential willmake it to the next phase, which is evaluation in pro-spective trials. Although the pipeline is most easily con-ceptualized as 4 distinct sequential steps, each phasecan involve multiple studies performed at various timepoints throughout the biomarker’s developmental life-cycle. Clearly, the process of biomarker developmentand validation is a multiphase procedure. It requirescollaboration between academia and industry, is time-consuming, and carries a large financial burden.
Biomarkers come in different forms, such as DNA,
RNA, proteins, peptides, and metabolites. Given thatproteins are the biological endpoints that control most
biological processes and can be quantified efficiently,economically, and with high analytical sensitivity in thecirculation, it is not surprising that proteins havegained the most attention as potential circulating bio-markers. With the discovery of soft ionization methodsby Fenn and Tanaka (who shared the 2002 Nobel Prizein Chemistry), the analysis of complex proteomic mix-tures using mass spectrometry became feasible. Sincethis discovery, thousands of studies have been per-formed focusing on deciphering the proteomes of var-ious biological materials, including blood, other fluids,tissues, and cell lines in the quest to identify novel bio-markers. As mass spectrometry–based proteomicsturned quantitative, the excitement regarding the po-tential to discover novel biomarkers was heightened(8). However, the number of biomarkers entering the
clinic has not increased, suggesting that the develop-ment pipeline is not efficient. A major problem is thatthe commonly used approach based on unbiased high-
Fig. 1. The biomarker development pipeline.
The 4 main phases of biomarker development are depicted as described in the text. For a biomarker to move from one phase
to another it needs to overcome multiple challenges at different levels. Only biomarkers that will reach the last step successfullywill be implemented in the clinic. Although depicted as 4 sequential steps, the phases are not always distinct from each otherand the pipeline is not always linear. Each phase can involve multiple studies performed at various time points throughout thebiomarker’s developmental lifecycle.Reviews
148 Clinical Chemistry 59:1 (2013)

throughput discovery to biomarker identification suf-
fers from relatively high false-positive rates. The largenumber of false positives, in turn, is slowing identifica-tion and validation of true biomarkers. Regrettably,even biomarkers with good performance never enterthe clinic, because of lack of rigorous validation or ascarcity of data showing a clear or superior contribu-tion to existing clinical practices. The key to tacklingboth these issues is to clearly define the clinical ques-tion that the biomarker should address before any bio-marker study is undertaken. If investigators do notknow the clinical question to be answered, it is impos-sible to design an appropriate study to identify and val-idate a true biomarker.
The journey of a protein biomarker from the
bench to the clinic is long and challenging, and everystep must be meticulously planned and executed tosucceed. This is evident not only from the numerousarticles addressing possible causes and potential solu-tions for the paucity of novel biomarkers, but also fromother documents such as the NIH Roadmap initiativeand the US FDA’s Critical Path Initiative for drugs anddiagnostics (9). Our goal in this review was to focus on
the clinical uses of protein cancer biomarkers, providingexamples and discussing their strengths and weaknesses.We mention barriers that a potential biomarker has toovercome to reach the clinic, suggest possible solutions tothese barriers, and summarize the regulatory proceduresrequired for a biomarker to obtain FDA approval. Finally,we follow the uphill path of a protein cancer biomarkerfrom the bench to the bedside.
Clinical Uses of Cancer BiomarkersIdeally, a cancer biomarker test would be a blood test
that is positive only in patients with cancer and is cor-related with disease stage, provides prognostic infor-mation, predicts response to treatment, and is easilyand reproducibly performed. To date, only 18 proteincancer biomarkers have been cleared by the FDA (Ta-ble 1). Although these are in clinical use, they are farfrom ideal. Below, we describe the main uses of cancerbiomarkers, provide examples, and highlight theirmain shortcomings.Table 1. FDA-cleared protein cancer biomarkers.
BiomarkerOfficial gene
nameaClinical use Cancer type Source type
/H9251-fetoprotein (AFP) AFP Staging Nonseminomatous testicular Serum
Human chorionic gonadotropin (hGC) CGB Staging Testicular Serum
Carbohydrate antigen 19-9 (CA19-9) Monitoring Pancreatic SerumCarbohydrate antigen 125 (CA125)
MUC16 Monitoring Ovarian Serum
Carcinoembryonic antigen (CEA) PSG2 Monitoring Colorectal Tissue
Epidermal growth factor receptor (EGFR) EGFR Prediction Colorectal Tissue
v-kit Hardy-Zuckerman 4 feline sarcoma viral
oncogene homolog (KIT)KIT Prediction Gastrointestinal Tissue
Thyroglobulin TG Monitoring Thyroid Serum
Prostate specific antigen (PSA) KLK3 Screening and monitoring Prostate Serum
Carbohydrate antigen 15.3 (CA 15.3) MUC1 Monitoring Breast Serum
Carbohydrate antigen 27.29 (CA27.29) MUC1 Monitoring Breast Serum
Estrogen receptor (ER) ESR1 Prognosis and prediction Breast Tissue
Progesterone receptor (PR) PGR Prognosis and prediction Breast Tissue
v-erb-b2 erythroblastic leukemia viral oncogene
homolog 2 (HER2-neu)ERBB2 Prognosis and prediction Breast Tissue
Nuclear matrix protein 22 (NMP-22) Screening and monitoring Bladder Urine
Fibrin/fibrinogen degradation products (FDP) Monitoring Bladder UrineBladder tumor antigen (BTA) Monitoring Bladder UrineHigh molecular CEA and mucin Monitoring Bladder Urine
aHuman genes: AFP, alpha-fetoprotein; CGB, chorionic gonadotropin, beta polypeptide; MUC16 , mucin 16, cell surface associated; PSG2 , pregnancy specific
beta-1-glycoprotein 2; EGFR , epidermal growth factor receptor; KIT, Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog; TG, thyroglobulin; KLK3 ,
kallikrein-related peptidase 3; MUC1 , mucin 1, cell surface associated; ESR1 , estrogen receptor 1; PGR, progesterone receptor; ERBB2 , v-erb-b2 erythroblastic
leukemia viral oncogene homolog 2, neuro/glioblastoma derived oncogene homolog (avian).Cancer Biomarkers from Bench to ClinicReviews
Clinical Chemistry 59:1 (2013) 149

SCREENING/EARLY DETECTION
“Early is the watchword for cancer control, early diag-
nosis, early treatment will save many lives” was the slo-gan on one of the posters from the Work Projects Ad-ministration Poster Collection (www.loc.gov/pictures/resource/cph.3b48900) designed to publicize healthand safety programs in the mid-1930s. But it was notuntil 1971, when President Richard Nixon declared thewar on cancer, that this concept was truly thrust intothe public spotlight. In most cases, by the time cancerbecomes symptomatic, the tumor has already spreadand treatments become ineffective. The objective of ascreening program is to detect cancer at a curable stage,before symptoms develop. For example, localizedbreast cancer has a 5-year survival rate of more than90%, whereas in patients with metastasis the 5-yearsurvival drops to 60% if regional and 30% if distant(10).
A useful screening biomarker must meet several
criteria. It should be able to detect the disease at an earlyand asymptomatic stage and result in decreased mor-bidity or increased survival rates. Accurately determin-ing the effect of early detection on survival rates can bedifficult and must factor in lead-time bias, the per-ceived increase in survival rates that occurs solely bydetecting the disease earlier. Moreover, a screening testmust be highly specific to minimize false positives. Inthis context, the term specificity refers to the propor-tion of healthy people with a marker result below es-tablished cutoffs. Within a population of 1 millionscreened individuals a test with 99.9% specificity willstill lead to 1000 false-positive results. Therefore, highspecificity is necessary for a good screening biomarkergiven that even a small false-positive rate could triggera large number of unnecessary diagnostic procedureswith the associated psychological stress and cost. Theoptimal sensitivity and specificity requirements for anymarker will depend on many factors and must take intoaccount the consequences of producing either a false-positive or false-negative result. Ultimately, for ascreening program to be considered successful, it mustbe cost-effective and noninvasive and lead to a measur-able reduction of disease-specific morbidity and/ormortality. As a result, screening for early detection isbest suited for diseases in which the prevalence is rela-tively high, medical care is accessible, and patients arewilling to collaborate for further follow-up and treat-ment (11).
One of the most well-known and studied cancer
screening biomarkers is prostate-specific antigen(PSA). In 1986 PSA was approved for the monitoringof prostate cancer, and this monitoring resulted in aconsiderable improvement in prostate cancer treat-ment. Eight years later, PSA was cleared by the FDA forscreening of prostate cancer, based on studies showingthat increased PSA concentrations in asymptomatic in-
dividuals were associated with increased risk of pros-tate cancer (12). Widespread implementation of the
PSA blood test resulted in early detection of approxi-mately 90% of prostate cancer cases and therefore con-tributed to an apparently longer survival period afterdiagnosis.
Despite its widespread use, PSA suffers from sev-
eral major limitations. In addition to being found inpatients with prostate cancer, increased concentrationsof PSA can be found in individuals with benign condi-tions such as benign prostate enlargement and prostateinflammation and infection. Additionally, PSA valuesdo not correlate well with tumor aggressiveness. High-grade prostate tumors display the greatest risk forspreading and lead to prostate cancer-related death,whereas low-grade tumors may remain localized andnot pose a threat to the patient’s life; treatment of thelatter type of tumors may be more harmful to patient’squality of life than the actual disease. As a result, thecontribution of PSA screening in decreasing disease-specific mortality is still being contested on the basis ofcontroversial results from 2 prospective prostatecancer–screening trials, the European RandomizedStudy of Screening for Prostate Cancer and the Pros-
tate, Lung, Colorectal, and Ovarian Cancer ScreeningTrial (13, 14) . Data from the European Randomized
Study of Screening for Prostate Cancer trial showedthat PSA screening resulted in a 20% decrease in theprostate cancer–specific mortality rate, whereas theProstate, Lung, Colorectal, and Ovarian CancerScreening Trial demonstrated no significant reductionin disease-specific mortality. Both trials showed thatPSA screening led to overdiagnosis and overtreatmentof prostate cancer, which were accompanied by poten-tially harmful results, including unnecessary biopsies,side effects from treatment, and increased psychologi-cal stress. Therefore, even PSA, the most widely knowncancer biomarker, is far from ideal. Thus a major un-met clinical need remains: the ability to screen for pros-tate cancer by using a method that has greater diagnos-tic specificity than PSA for prostate cancer and candiscriminate between indolent and aggressive (poten-tially life-threatening) disease.
DIAGNOSIS
A diagnostic test, contrary to screening, would be pre-scribed to an individual who has presented with symp-toms. However, the characteristics of an ideal diagnos-tic biomarker are similar to the characteristics forscreening. Notably, there is no protein biomarker rec-ommended in practice guidelines for cancer diagnosis,but many of the well-known markers are widely used asdiagnostic aids.Reviews
150 Clinical Chemistry 59:1 (2013)

PREDICTION AND PROGNOSIS
A prognostic factor is a patient or disease characteristic
at the time of diagnosis that provides informationabout the natural history of the disease, independent oftherapy. On the other hand, a predictive biomarkerpredicts response to a therapeutic intervention com-pared to the effect of a different therapy or no treat-ment. Predictive biomarkers form the foundations forpersonalized medicine.
A prognostic test is used to classify patients into
different risk groups. If the prognostic value of certainbiomarkers is high, these markers may eventually beincorporated into the tumor-node-metastasis (TNM)staging system, as is the case for 3 serum biomarkers,
/H9251
fetoprotein, human chorionic gonadotropin, and lac-tate dehydrogenase in testicular cancer, for which thereis a TNMS system in place with a site-specific prognos-tic factor (S is for site-specific prognostic factors) (15).
Prognostic biomarkers are important because patientswith a good prognosis could be spared from receivingadjuvant therapy, thus avoiding side effects and reduc-ing treatment cost. Because most prognostic biomark-ers lack the required diagnostic accuracy, cliniciansprefer overtreatment to undertreatment.
Evaluating prognostic and predictive biomarkers,
especially in a prospective manner, is challenging giventhat endpoints of interest such as disease-free and over-all survival take years to reach and often require com-plicated statistical analyses (16).
The most widely used prognostic and predictive
biomarker is the estrogen receptor (ER) tissue markerin breast cancer. For prognosis, patients with ER-positive tumors have better outcomes those with ER-negative tumors. However, the prognostic value of ERis time dependent (5 years or earlier after diagnosis),and the prognostic impact of ER in lymph-node–negative patients is limited (17, 18) . The need for more
accurate prognostic breast cancer biomarkers has led tointensified research and resulted in a series of multi-gene classifiers, proposed as potentially useful adjunctsfor breast cancer patient management. However, theclinical value of these classifiers remains to be estab-lished (19).
The ER was the first predictive biomarker recom-
mended for routine use in breast cancer by the TumorMarker Panel of the American Society of Clinical On-cology (20). It is used for selecting patients likely to
respond to hormonal therapy. Endocrine treatment,such as selective ER modulators and aromatase inhib-itors, are the most effective in patients with hormonereceptor–positive tumors in early or advanced disease(21). However, the predictive efficacy of ER is far from
ideal, because approximately one third of advancedER-positive patients are intrinsically resistant to endo-crine therapies, and the majority of ER-positive tumorsthat initially respond to endocrine therapy will eventu-
ally develop resistance (22).
MONITORING
Following therapy, cancer patients are monitored to
ensure that they remain disease free or are treatedpromptly upon relapse. Additionally, monitoring bio-marker concentrations could be indicative of a thera-peutic response, with increasing concentrations associ-ated with resistance and consideration for alternativetherapies. A monitoring test needs to be both diagnos-tically sensitive and specific to ensure continuation ofbeneficial therapies and early discontinuation/replace-ment of ineffective therapies.
A typical example of a monitoring biomarker is
the carbohydrate antigen 19-9 (CA19-9). CA19-9 wasidentified in 1979 as a useful marker for pancreatic andcolorectal cancer (23) and in 2002 it was cleared by the
FDA for monitoring pancreatic cancer patients.CA19-9 has relatively good diagnostic sensitivity(70%–90% in advance disease, approximately 50% in
early disease, and absent in approximately 5% of thegeneral population) but poor diagnostic specificity(increased in numerous other cancer types as well asin benign pancreatic diseases and hepatobiliary in-flammation) (24–27) . Therefore, CA19-9 is recom-
mended to assess response to therapy in patientswith increased CA19-9 concentrations before treat-ment but is not recommended as a screening markeror as the sole method to identify recurrence of pan-creatic cancer (28) .
The Need for New Cancer BiomarkersThere are 18 FDA-cleared protein cancer biomarkers
(Table 1), with several others being used clinicallywithout FDA clearance. There is a clear need to identifyadditional biomarkers for optimal patient manage-ment. Numerous cancer researchers and organizationshave strived to identify novel biomarkers that can fulfillthese unmet clinical needs. So why are there so few newbiomarkers entering the clinic, and can anything bedone about it?
Challenges to Biomarker DevelopmentBelow, we outline some challenges of the biomarker-
development pipeline after the discovery phase andsuggest some ways to overcome these challenges. Theseare also summarized in Table 2. We also describe thepath of a prognostic biomarker, which is an examplethat has yet to be cleared by the FDA despite over 20years of testing and a wealth of supportive data.Cancer Biomarkers from Bench to ClinicReviews
Clinical Chemistry 59:1 (2013) 151

ASSAY
A clinically useful biomarker should be measured reli-
ably, so ease of adoption for use by routine clinicallaboratories is key (29). Immunoassays such as ELISA
remain the gold standard for validation and clinical useof protein biomarkers (30, 31) . Mass spectrometry–
based approaches for clinical applications remain anappealing prospect for numerous reasons including,but not limited to, high analytical specificity and sensi-tivity and multiplexing capabilities. However, assaycomplexity and expertise requirements are currentlypreventing the adoption of mass spectrometry intoroutine use in clinical laboratories (30).
Before assay development, numerous preanalyti-
cal variables must be assessed, such as choice of a bio-logical matrix and sample collection, handling, andstorage. Depending on the availability of critical re-agents (antibodies, calibrators), assay development cantake from weeks to years. In the case of immunoassays,the lack of high-quality antibodies has slowed the rateof development. To abbreviate that barrier, the HumanProtein Atlas project took the initiative of developingmultiple antibodies against all human proteins (32).A s
of November 2011, 15 598 antibodies corresponding to12 238 protein-coding genes that cover /H1102240% of the
19 559 human entries as defined by UniProt have beengenerated (33). The generated prototype assay needs
careful validation. Basic analytical characteristics to beexamined include assay dose–response curve, measur-ing range, limits of detection and quantification, accu-racy, imprecision, and analytical specificity (34). Ad-
hering to federal regulations that govern humandiagnostic testing, known as the Clinical LaboratoryImprovement Amendments, during analytical assayvalidation could increase the reliability and quality ofthe data (35).
The importance of assay technical quality was un-
derpinned in a recent study in which tissue samplescollected in the Prostate, Lung, Colorectal and OvarianCancer Screening Trial were used to evaluate potentialovarian cancer biomarkers (36). In this study prediag-
nostic samples were used to evaluate in a phase III set-ting the performance of 28 ovarian cancer biomarkersthat have shown promise in phase II studies. Althoughthe main conclusion was that CA125 outperformed allother biomarkers, one finding was that assay perfor-mance correlates strongly with biomarker perfor-mance. None of the markers for which the assays hadTable 2. Reasons for biomarker failures.
Reason Solutions Frequency Examples
Fraud Low Potti et al. (79 )
Preanalytical factors  Clearly define the clinical question to be addressed High Xu et al. (80 )
 Patient selection bias  Use samples collected under detailed SOPs Villanueva et al. (81 )
 Sample collection, handling and
storage Use well annotated samples
Analytical factors  Validate the analytical method High Leman et al. (82 )
 Methodological artefacts  Use appropriate quality controls with all analyses
 Poor analytical method
Statistics/Bioinformatics  Seek and follow the expertise of an experienced
biostatisticianHigh Mor et al. (83 )
Petricoin et al. (84 ) Inappropriate statistical analysis
 Data overfitting Small sample size Multiple hypothesis testing Overlapping training and validation
patient cohorts
Clinical validation
 Nonreproducible validation Poor study design No adequate clinical performance Clearly define the clinical question to be addressed
prior to undertaking any study
 Collaborate with an experienced biostatistician Use appropriate specimens to avoid bias Use validated analytical methodsMedium Esrig et al.
(85 )
Malats et al. (86 )
Kim et al. (87 )
Commercialization  Apply for patents to obtain intellectual property
rights as early as possibleLow
 Intellectual property
FDA approval  Seek FDA guidance early in development phases LowReviews
152 Clinical Chemistry 59:1 (2013)

CVs/H1102230% had adequate diagnostic sensitivity in
phase II or phase III studies.
SAMPLE REQUIREMENTS
Identifying the appropriate target population to ad-dress the clinical question to be answered and ensuringsample integrity from collection to analysis are essen-tial steps toward biomarker development. The popula-tion used in the validation, including both cases andcontrols, should be selected to closely match the targetpopulation, and the selection process must have clearlydefined inclusion/exclusion criteria. Sample size re-quirements must be calculated to ensure adequate sta-tistical power for each study (37). Sample sizes usually
increase as the biomarker moves toward the clinic.Patient-related factors, including fasting, posture, cir-cadian rhythms, age, and sex should be documentedfor every sample and carefully investigated to exploretheir relation to the analyte of interest. Any factorfound to affect analyte concentrations must be thor-oughly controlled in all future studies. Additionally,biomarker stability should be studied under differentconditions to identify the most efficient protocol forcollection, handling, and storage. Finally, extra cau-
tion should be given to devices and reagents usedduring sample collection and storage. Collection andprocessing of all samples should be performed usingpredefined standard operating procedures to mini-mize preanalytical biases. Using samples that arepoorly annotated or from an inappropriate popula-tion will introduce bias into studies and lead to falseresults.
Access to desired biological resources has been one
of the limiting factors during validation studies, andthe need for biological samples far surpasses the avail-able supply. High-quality samples with associated clin-ical data are sparse and “should be reserved for the
most promising research studies or late-stage bio-marker discovery efforts” (38). Toward this direction,
there are efforts to create a central registry that willinclude information on novel biomarkers undergoingvalidation studies, especially those that are using sam-ples from randomized trials, so that the most promis-ing candidates can be identified for further study(39) . In addition, efforts are also underway to gen-
erate high-quality sample collections for validatingpromising biomarkers. A large NIH initiative, theEarly Detection Research Network, promotes col-laboration among academic and industrial research-ers and organizations, both nationally and interna-tionally, for the development and testing ofpromising biomarkers or technologies for early de-tection of cancer (http://edrn.nci.nih.gov).STATISTICS
Robust statistical analysis of validation data of poten-
tial biomarkers is essential. Possible biases, includingsmall sample size, inappropriate controls, noninde-pendent training and validation cohorts, and multiplehypothesis testing should be critically examined. Fur-thermore, statistical significance, frequently repre-sented by Pvalues, is not adequate to assess clinical
utility and can even be misleading. Researchers shouldclosely collaborate with biostatisticians that are experi-enced in the biomarker field to help plan and performtheir study.
With the advent of -omics technologies such as
RNA microarrays and the development of multigenesignatures, statistical analysis of validation studies be-came more complex. Initially, the statistical methodsused for analyzing complex outputs relied on tests de-veloped and optimized for single-signal output. How-ever, undertaken efforts resulted in newer and moreappropriate methodologies, based on multisignal read-ings (40). For example, the output of multigene and
multiprotein diagnostic tools, such as Oncotype Dx(41) and OVA1 (42), is a numeric score calculated
from expression values of multiple genes or proteins,respectively. Although the algorithm used to calculatethe diagnostic score is a vital component, it remains a“black box” for clinicians. To address these types oftests the FDA created a new diagnostic category entitled“in vitro diagnostic multivariate index assay” (43).
FUNDING
As a biomarker moves from one phase of development
to another, the financial burden increases. Soon afterdiscovery and early validation studies, usually takingplace in an academic setting, industry may enter thescene and provide financial support for the remainingdevelopment phases. Being able to clearly demonstratethat a biomarker addresses an unmet clinical need bycarefully planning and executing early validation stud-ies is critical to acquiring an industry partner. A poten-tial limitation at that stage may be issues related to in-tellectual property. A key element of biomarkercommercialization is ownership, and industrial part-ners will not undertake the development of a markerthat does not have potential for investment return, re-gardless of its clinical utility. Therefore investigatorsshould apply for intellectual property rights as early aspossible in the biomarker discovery process.
One of the reasons to seek FDA clearance of a bio-
marker is to obtain reimbursement by Medicare or pri-vate insurers, allowing for broader use of the test. Onthe other hand, inadequate reimbursement rates maydiscourage industrial partners from seeking FDA ap-proval, opting for development of an in-house testwhich may receive less utilization but with higher re-Cancer Biomarkers from Bench to ClinicReviews
Clinical Chemistry 59:1 (2013) 153

imbursement rates. The importance of inadequate re-
imbursement is highlighted by the existence of 2government-commissioned reports recommendingthe reevaluation of reimbursement rates for diagnos-tics(44, 45) . Because of their high financial risks, diag-
nostics have been regarded by industry as a less attrac-tive investment opportunity than therapeutics, andconsequently the development of diagnostic methodsis not as well funded by industry as the development oftherapeutic drugs (46). This reluctance to invest in di-
agnostics poses a challenge for biomarker developmentand further highlights the importance of carefully de-fining the clinical question to be addressed by the bio-marker before initiating any biomarker discoveryefforts.
REGULATORY ISSUES
Clinical validation of novel potential cancer biomark-ers includes assessing their diagnostic utility in studieswith different populations (37). After potential clinical
usefulness is demonstrated, a company may pursue de-velopment of a product (in vitro diagnostic device) andconduct further validation studies to evaluate the testtechnically and clinically. When sufficient data areavailable on clinical utility for a specified clinical use,approval from regulatory agencies such as the FDAmay be sought. Although obtaining regulatory ap-proval is not necessary to market or obtain reimburse-ment for diagnostic tests in the US, obtaining such ap-proval can increase the testing market and allow themarker to gain wider acceptance.
The FDA has established the Voluntary Explor-
atory Data Submission, which encourages sponsors toshare their data with the agency, even at early stages ofthe development, without being considered as part ofthe regulatory decision-making process. The agencyleans toward establishing a collaborative relationshipwith the sponsors rather than serving strictly as a reg-ulatory body.
More details regarding regulatory processes for
biomarker development in the US can be found on theFDA website (www.fda.gov) as well as in recent reviews(3, 47) . Every country has its own regulatory body that
governs the requirements for biomarker use within itsown medical system, and approval by one country’sregulatory body does not translate into approval byanother.
The Example of Urokinase-Type Plasminogen
Activator and Its Inhibitor Plasminogen ActivatorInhibitor Type 1
We will use a biomarker with proven clinical utility, but
not widespread use, to outline some major implemen-tation difficulties with cancer biomarkers. As men-tioned earlier, one of the unmet clinical needs in breast
cancer is the development of more diagnostically sen-sitive and specific prognostic markers (48). Current
breast cancer prognosticators include lymph node sta-tus; number of involved nodes; tumor size, grade, andhormone receptor status; and HER2 (human epider-mal growth factor receptor 2) amplification (49, 50) .
However, these factors can provide unequivocal prog-nostic information (favorable or poor) for only 30% ofbreast cancer patients (10).
The serine protease urokinase-type plasminogen
activator (uPA) and its inhibitor, plasminogen activa-tor inhibitor type 1 (PAI-1), is one example of potentialprognostic breast cancer biomarkers. The concentra-tions of uPA and PAI-1 antigen correlate with recur-rence risk; patients with node-negative breast cancerand low concentrations of uPA and PAI-1 can be clas-sified as a low-risk subgroup and thus be spared sys-temic adjuvant therapy (51).
The prognostic potential of uPA and PAI-1 is not a
recent finding. In 1988 Duffy et al. published a prelim-inary report demonstrating the prognostic potential ofthe enzymatic activity of serine protease uPA in pa-tients with primary breast cancer (52). In less than a
year, Janicke et al. reported that the tumor tissue anti-gen concentration of uPA (apart from its activity) alsohas prognostic relevance in breast cancer (53).I nt h e
following 3 years, PAI-1 also emerged as an additionalprognostic breast cancer biomarker in both node-negative and node-positive breast cancer patients (54).
Subsequently, between the early 1990s and mid-2000s,15 individual studies (55–67) covering a variety of de-
mographic areas supported the prognostic impact ofuPA and PAI-1 in primary breast cancer.
The ELISAs used for determining antigen concen-
trations of uPA and PAI-1 were assessed by the Recep-tor and Biomarker Group of the European Organiza-tion for Research and Treatment of Cancer, andobtained international quality assurance (68, 69) . Ad-
ditionally, the optimum method for tissue preparationbefore analysis was determined, further standardizingthe measurement of these markers (70).
uPA and PAI-1 are the first tumor markers to
reach the highest level of evidence validation for theirclinical utility in breast cancer management. The high-est level of evidence, as defined by Hayes et al., can beachieved through “evidence from a single, high-powered, prospective, controlled study that is specifi-cally designed to test the marker, or evidence frommetaanalysis and/or overview of level II or III studies”(71). In the case of uPA/PAI-1, a metaanalysis of 18
datasets encompassing more than 8000 primary breastcancer patients with a median follow-up of 6.5 yearswas performed by the Receptor and Biomarker Groupof the European Organization for Research and Treat-Reviews
154 Clinical Chemistry 59:1 (2013)

ment of Cancer (51) and verified the prognostic poten-
tial of the markers. Additionally, a prospective ran-domized multicenter therapy trial (Chemo N
0) that
was performed during 1993–1999 confirmed the prog-
nostic impact of uPA and PAI-1 (72).
Taking everything into account, more than 20
years of intense research and collaboration between ac-ademia, industry, physicians, and patients were re-quired for demonstrating the clinical utility of these 2protein cancer markers. Of note, uPA and PAI-1 arefound in the list of recommended prognostic tumormarkers for breast cancer published by the AmericanSociety of Clinical Oncology (73); yet they are still not
FDA cleared.
On the basis of this example, one could conclude
that, even in the case of biomarkers with clear clinicalutility that are reproducible in independent studies,have analytically verified assays, and have achievedhigh levels of evidence for clinical validation, the pro-cess from bench to the clinic could last 2 or more de-cades. A possible major drawback in the use of uPA/PAI is the requirement for fresh-frozen tissue, a type ofspecimen that is not collected on a regular basis. Even ifall the essential components of success are present (truediscovery, robust clinical assay, clinical trials, funding),biomarker clinical implementation requires time, andthere is no easy way to bypass that constraint. In gen-eral, the greater the clinical impact the marker has, theless time it will take to be adopted for use in the clinic.Human epididymis protein 4, for example, was firstreported as a potential ovarian cancer biomarker in thelate 1990s (74–76) and obtained FDA approval for
monitoring recurrence and progression of ovariancancer in 2010. PSA took approximately 6 years fromits first detection in serum (77, 78) to FDA clearance
for monitoring and another 8 years for its clearance forscreening.
ConclusionThe clinical uses of cancer biomarkers are quite diverse.
The intensified research for developing novel biomark-ers stems from the fact that markers currently in clini-
cal practice suffer from key limitations. Despite thehigh expectations regarding the availability of informa-tion from the completed human genome sequencingproject and the advent of promising technological ad-vances, the number of novel protein cancer biomarkersobtaining FDA clearance has decreased during the lastdecade. Many see this as a failure of the biomarker and-omics field to deliver results. However, we need to
consider that biomarker development encompassesmultiple contiguous phases, requires collaborationamong different stakeholders, carries a major financialburden and faces many potential challenges. Evenwhen all of these challenges are overcome, the historyof clinically useful biomarkers suggests that at least adecade is required for the transition of a marker frombench to the bedside. Therefore, it may be too early toexpect that the new technological advances will cata-lyze the anticipated biomarker revolution any timesoon.
Author Contributions: All authors confirmed they have contributed
to the intellectual content of this paper and have met the following 3requirements: (a) significant contributions to the conception and design,acquisition of data, or analysis and interpretation of data; (b) draftingor revising the article for intellectual content; and (c) final approval ofthe published article.
Authors’ Disclosures or Potential Conflicts of Interest: Upon man-
uscript submission, all authors completed the author disclosure form.Disclosures and/or potential conflicts of interest:
Employment or Leadership: E.P. Diamandis, Clinical Chemistry ,
AACC.Consultant or Advisory Role: None declared.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: None declared.
Expert Testimony: None declared.
Patents: None declared.
Role of Sponsor: The funding organizations played no role in the
design of study, choice of enrolled patients, review and interpretationof data, or preparation or approval of manuscript.
References
1.Biomarkers Definitions Working Group. Biomark-
ers and surrogate endpoints: preferred definitionsand conceptual framework. Clin Pharmacol Ther2001;69:89 –95.
2.Anderson NL, Anderson NG. The human plasma
proteome: history, character, and diagnostic pros-pects. Mol Cell Proteomics 2002;1:845– 67.
3.Gutman S, Kessler LG. The US Food and Drug
Administration perspective on cancer biomarkerdevelopment. Nat Rev Cancer 2006;6:565–71.
4.Kulasingam V, Diamandis EP. Strategies for dis-
covering novel cancer biomarkers through utili-zation of emerging technologies. Nat Clin PractOncol 2008;5:588 –99.
5.deVera IE, Katz JE, Agus DB. Clinical proteo-
mics: the promises and challenges of massspectrometry-based biomarker discovery. ClinAdv Hematol Oncol 2006;4:541–9.
6.Rifai N, Gillette MA, Carr SA. Protein biomarker
discovery and validation: the long and uncertainpath to clinical utility. Nat Biotechnol 2006;24:971– 83.
7.Wang P, Whiteaker JR, Paulovich AG. The evolv-
ing role of mass spectrometry in cancer biomarkerdiscovery. Cancer Biol Ther 2009;8:1083–94.
8.Ong SE, Mann M. Mass spectrometry-based pro-teomics turns quantitative. Nat Chem Biol 2005;
1:252– 62.
9.US Department of Health and Human Services,
Food and Drug Administration. Innovation orstagnation: challenge and opportunity on the crit-
ical path to new medical products. 2004. http://
www.fda.gov/ScienceResearch/SpecialTopics/CriticalPathInitiative/CriticalPathOpportunitiesReports/ucm077262.htm (Accessed December2012).
10. Weigelt B, Peterse JL, van ’t Veer LJ. Breast
cancer metastasis: markers and models. Nat RevCancer 2005;5:591– 602.Cancer Biomarkers from Bench to ClinicReviews
Clinical Chemistry 59:1 (2013) 155

11. Mulley A. Primary care medicine. 3rd ed.
Philadelphia: J.B. Lippincott; 1995: p 13– 6.
12. Catalona WJ, Smith DS, Ratliff TL, Dodds KM,
Coplen DE, Yuan JJ, et al. Measurement ofprostate-specific antigen in serum as a screeningtest for prostate cancer. N Engl J Med 1991;324:1156 – 61.
13. Andriole GL, Crawford ED, Grubb RL, III, Buys SS,
Chia D, Church TR et al. Mortality results from arandomized prostate-cancer screening trial.N Engl J Med 2009;360:1310 –9.
14. Schroder FH, Hugosson J, Roobol MJ, Tammela
TL, Ciatto S, Nelen V, et al. Screening andprostate-cancer mortality in a randomized Euro-pean study. N Engl J Med 2009;360:1320 – 8.
15. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene
FL, Trotti A. Testis. AJCC cancer staging manual.New York: Springer Verlag; 2009. p 469 –73.
16. Buyse M, Sargent DJ, Grothey A, Matheson A, de
Gramont A. Biomarkers and surrogate end points:the challenge of statistical validation. Nat RevClin Oncol 2010;7:309 –17.
17. Duffy MJ. Biochemical markers as prognostic in-
dices in breast cancer. Clin Chem 1990;36:188 –91.
18. Isaacs C, Stearns V, Hayes DF. New prognostic
factors for breast cancer recurrence. Semin Oncol2001;28:53– 67.
19. Ross JS, Hatzis C, Symmans WF, Pusztai L, Hor-
tobagyi GN. Commercialized multigene predictorsof clinical outcome for breast cancer. Oncologist2008;13:477–93.
20. ASCO Tumor Marker Expert Panel. Clinical prac-
tice guidelines for the use of tumor markers inbreast and colorectal cancer. Adopted on May 17,1996 by the American Society of Clinical Oncol-ogy. J Clin Oncol 1996;14:2843–77.
21. Duffy MJ. Predictive markers in breast and other
cancers: a review. Clin Chem 2005;51:494 –503.
22. Hanstein B, Djahansouzi S, Dall P, Beckmann
MW, Bender HG. Insights into the molecular bi-ology of the estrogen receptor define novel ther-apeutic targets for breast cancer. Eur J Endocrinol2004;150:243–55.
23. Koprowski H, Steplewski Z, Mitchell K, Herlyn M,
Herlyn D, Fuhrer P. Colorectal carcinoma antigensdetected by hybridoma antibodies. Somatic CellGenet 1979;5:957–71.
24. Tempero MA, Uchida E, Takasaki H, Burnett DA,
Steplewski Z, Pour PM. Relationship of carbohy-drate antigen 19-9 and Lewis antigens in pancre-atic cancer. Cancer Res 1987;47:5501–3.
25. Goonetilleke KS, Siriwardena AK. Systematic re-
view of carbohydrate antigen (CA 19-9) as abiochemical marker in the diagnosis of pancreaticcancer. Eur J Surg Oncol 2007;33:266 –70.
26. Frebourg T, Bercoff E, Manchon N, Senant J,
Basuyau JP, Breton P, et al. The evaluation of CA19-9 antigen level in the early detection of pan-creatic cancer. A prospective study of 866 pa-tients. Cancer 1988;62:2287–90.
27. Rosty C, Goggins M. Early detection of pancreatic
carcinoma. Hematol Oncol Clin North Am 2002;16:37–52.
28. Locker GY, Hamilton S, Harris J, Jessup JM, Ke-
meny N, Macdonald JS, et al. ASCO 2006 updateof recommendations for the use of tumor markersin gastrointestinal cancer. J Clin Oncol 2006;24:5313–27.
29. US Department of Health and Human Services,Food and Drug Administration. Guidance for
industry: bioanalytical method validation. 2001.http://www.fda.gov/downloads/Drugs/…/Guidances/ucm070107.pdf (Accessed December2012).
30. Sahab ZJ, Semaan SM, Sang QX. Methodology
and applications of disease biomarker identifica-tion in human serum. Biomark Insights 2007;2:21– 43.
31. Veenstra TD. Global and targeted quantitative
proteomics for biomarker discovery. J Chro-matogr B Analyt Technol Biomed Life Sci 2007;847:3–11.
32. Uhlen M, Bjorling E, Agaton C, Szigyarto CA,
Amini B, Andersen E, et al. A human protein atlasfor normal and cancer tissues based on antibodyproteomics. Mol Cell Proteomics 2005;4:1920 –32.
33. Uhlen M, Oksvold P, Fagerberg L, Lundberg E,
Jonasson K, Forsberg M, et al. Towards aknowledge-based human protein atlas. Nat Bio-technol 2010;28:1248 –50.
34. Lee JW, Hall M. Method validation of protein
biomarkers in support of drug development orclinical diagnosis/prognosis. J Chromatogr B Ana-lyt Technol Biomed Life Sci 2009;877:1259 –71.
35. Swanson BN. Delivery of high-quality biomarker
assays. Dis Markers 2002;18:47–56.
36. Cramer DW, Bast RC Jr, Berg CD, Diamandis EP,
Godwin AK, Hartge, P et al. Ovarian cancer bio-marker performance in prostate, lung, colorectal,and ovarian cancer screening trial specimens.Cancer Prev Res 2011;4:365–74.
37. Pepe MS, Etzioni R, Feng Z, Potter JD, Thompson
ML, Thornquist M, et al. Phases of biomarkerdevelopment for early detection of cancer. J NatlCancer Inst 2001;93:1054 – 61.
38. Hinestrosa MC, Dickersin K, Klein P, Mayer M,
Noss K, Slamon D, et al. Shaping the future of
biomarker research in breast cancer to ensureclinical relevance. Nat Rev Cancer 2007;7:309 –15.
39. Andre F, McShane LM, Michiels S, Ransohoff DF,
Altman DG, Reis-Filho JS, et al. Biomarkerstudies: a call for a comprehensive biomarkerstudy registry. Nat Rev Clin Oncol 2011;8:171– 6.
40. US Department of Health and Human Services, Food
and Drug Administration. Statistical guidance on re-porting results from studies evaluating diagnostic tests.2007. http://www.fda.gov/MedicalDevices/DeviceRegulationandGuidance/GuidanceDocuments/ucm071148.htm (Accessed December 2012).
41. Cobleigh MA, Tabesh B, Bitterman P, Baker J,
Cronin M, Liu ML, et al. Tumor gene expressionand prognosis in breast cancer patients with 10or more positive lymph nodes. Clin Cancer Res2005;11:8623–31.
42. Zhang Z, Chan DW. The road from discovery to
clinical diagnostics: lessons learned from the firstFDA-cleared in vitro diagnostic multivariate indexassay of proteomic biomarkers. Cancer EpidemiolBiomarkers Prev 2010;19:2995–9.
43. US Department of Health and Human Services, Food
and Drug Administration. Draft guidance for indus-try, clinical laboratories, and FDA staff: in vitrodiagnostic multivariate index assays. 2007. http://www.fda.gov/downloads/MedicalDevices/DeviceRegulationandGuidance/GuidanceDocuments/ucm071455.pdf (Accessed December2012).44. Department of Health and Human Services, Secretary’s
Advisory Committee on Genetics, Health, and Society.Coverage and reimbursement of genetic tests andservices: report of the Secretary’s Advisory Committeeon Genetics, Health, and Society. 2006. http://oba.od.nih.gov/oba/sacghs/reports/CR_report.pdf (Ac-cessed December 2012).
45. Kessler L, Ramsey SD, Tunis S, Sullivan SD. Clin-
ical use of medical devices in the “BermudaTriangle.” Health Aff 2004;23:200 –7.
46. Phillips KA, Van BS, Issa AM. Diagnostics and
biomarker development: priming the pipeline.Nat Rev Drug Discov 2006;5:463–9.
47. Scherf U, Becker R, Chan M, Hojvat S. Approval of
novel biomarkers: FDA’s perspective and majorrequests. Scand J Clin Lab Invest Suppl 2010;242:96 –102.
48. Gradishar WJ. The future of breast cancer: the
role of prognostic factors. Breast Cancer ResTreat 2005;89(Suppl 1):S17–S26.
49. Pinder SE, Ellis IO, Galea M, O’Rouke S, Blamey
RW, Elston CW. Pathological prognostic factors in
breast cancer, III. Vascular invasion: relationshipwith recurrence and survival in a large study withlong-term follow-up. Histopathology 1994;24:41–7.
50. Rosen PP, Groshen S, Saigo PE, Kinne DW, Hell-
man S. Pathological prognostic factors in stage I(T1N0M0) and stage II (T1N1M0) breastcarcinoma: a study of 644 patients with medianfollow-up of 18 years. J Clin Oncol 1989;7:1239 –51.
51. Look MP, van Putten WL, Duffy MJ, Harbeck N,
Christensen IJ, Thomssen C, et al. Pooled analysisof prognostic impact of urokinase-type plasmin-ogen activator and its inhibitor PAI-1 in 8377breast cancer patients. J Natl Cancer Inst 2002;94:116 –28.
52. Duffy MJ, O’Grady P, Devaney D, O’Siorain L,
Fennelly JJ, Lijnen HJ. Urokinase-plasminogen ac-
tivator, a marker for aggressive breast carcino-mas. Preliminary report. Cancer 1988;62:531–3.
53. Janicke F, Schmitt M, Ulm K, Gossner W, Graeff
H. Urokinase-type plasminogen activator antigenand early relapse in breast cancer. Lancet 1989;2:1049.
54. Janicke F, Schmitt M, Graeff H. Clinical relevance
of the urokinase-type and tissue-type plasmino-gen activators and of their type 1 inhibitor inbreast cancer. Semin Thromb Hemost 1991;17:303–12.
55. Janicke F, Schmitt M, Pache L, Ulm K, Harbeck N,
Hofler H, Graeff H. Urokinase (uPA) and its inhib-itor PAI-1 are strong and independent prognosticfactors in node-negative breast cancer. BreastCancer Res Treat 1993;24:195–208.
56. Foekens JA, Schmitt M, van Putten WL, Peters
HA, Kramer MD, Janicke F, Klijn JG. Plasminogenactivator inhibitor-1 and prognosis in primarybreast cancer. J Clin Oncol 1994;12:1648 –58.
57. Grondahl-Hansen J, Peters HA, van Putten WL,
Look MP, Pappot H, Ronne E, et al. Prognosticsignificance of the receptor for urokinase plas-minogen activator in breast cancer. Clin CancerRes 1995;1:1079 – 87.
58. Eppenberger U, Kueng W, Schlaeppi JM, Roesel
JL, Benz C, Mueller H, et al. Markers of tumorangiogenesis and proteolysis independently de-fine high- and low-risk subsets of node-negativebreast cancer patients. J Clin Oncol 1998;16:3129 –36.
Reviews
156 Clinical Chemistry 59:1 (2013)

59. Kim SJ, Shiba E, Kobayashi T, Yayoi E, Furukawa
J, Takatsuka Y, et al. Prognostic impact ofurokinase-type plasminogen activator (PA), PAinhibitor type-1, and tissue-type PA antigen levelsin node-negative breast cancer: a prospectivestudy on multicenter basis. Clin Cancer Res 1998;4:177– 82.
60. Knoop A, Andreasen PA, Andersen JA, Hansen S,
Laenkholm AV, Simonsen AC, et al. Prognosticsignificance of urokinase-type plasminogen acti-vator and plasminogen activator inhibitor-1 inprimary breast cancer. Br J Cancer 1998;77:932–40.
61. Bouchet C, Hacene K, Martin PM, Becette V,
Tubiana-Hulin M, Lasry S, et al. Disseminationrisk index based on plasminogen activator systemcomponents in primary breast cancer. J Clin Oncol1999;17:3048 –57.
62. Foekens JA, Peters HA, Look MP, Portengen H,
Schmitt M, Kramer MD, et al. The urokinasesystem of plasminogen activation and prognosisin 2780 breast cancer patients. Cancer Res 2000;60:636 – 43.
63. Konecny G, Untch M, Arboleda J, Wilson C, Kahl-
ert S, Boettcher B, et al. Her-2/neu and urokinase-type plasminogen activator and its inhibitor inbreast cancer. Clin Cancer Res 2001;7:2448 –57.
64. Cufer T, Borstnar S, Vrhovec I. Prognostic and
predictive value of the urokinase-type plasmino-gen activator (uPA) and its inhibitors PAI-1 andPAI-2 in operable breast cancer. Int J Biol Markers2003;18:106 –15.
65. Zemzoum I, Kates RE, Ross JS, Dettmar P, Dutta
M, Henrichs C, et al. Invasion factors uPA/PAI-1and HER2 status provide independent and com-plementary information on patient outcome innode-negative breast cancer. J Clin Oncol 2003;21:1022– 8.
66. Bouchet C, Ferrero-Pous M, Hacene K, Becette V,
Spyratos F. Limited prognostic value of c-erbB-2compared to uPA and PAI-1 in primary breastcarcinoma. Int J Biol Markers 2003;18:207–17.
67. Meo S, Dittadi R, Peloso L, Gion M. The prognos-
tic value of vascular endothelial growth factor,urokinase plasminogen activator and plasmino-gen activator inhibitor-1 in node-negative breastcancer. Int J Biol Markers 2004;19:282– 8.
68. Sweep CG, Geurts-Moespot J, Grebenschikov N,
de Witte JH, Heuvel JJ, Schmitt M, et al. Externalquality assessment of trans-European multicentre
antigen determinations (enzyme-linked immu-nosorbent assay) of urokinase-type plasminogenactivator (uPA) and its type 1 inhibitor (PAI-1) inhuman breast cancer tissue extracts. Br J Cancer1998;78:1434 – 41.
69. Sweep FC, Fritsche HA, Gion M, Klee GG, Schmitt
M. Considerations on development, validation,application, and quality control of immuno(met-ric) biomarker assays in clinical cancer research:an EORTC-NCI working group report. Int J Oncol2003;23:1715–26.
70. Janicke F, Pache L, Schmitt M, Ulm K, Thomssen
C, Prechtl A, Graeff H. Both the cytosols anddetergent extracts of breast cancer tissues aresuited to evaluate the prognostic impact of theurokinase-type plasminogen activator and its in-hibitor, plasminogen activator inhibitor type 1.Cancer Res 1994;54:2527–30.
71. Hayes DF, Bast RC, Desch CE, Fritsche H Jr,
Kemeny NE, Jessup JM, et al. Tumor markerutility grading system: a framework to evaluateclinical utility of tumor markers. J Natl Cancer Inst1996;88:1456 – 66.
72. Janicke F, Prechtl A, Thomssen C, Harbeck N,
Meisner C, Untch M, et al. Randomized adjuvantchemotherapy trial in high-risk, lymph node-negative breast cancer patients identified byurokinase-type plasminogen activator and plas-minogen activator inhibitor type 1. J Natl CancerInst 2001;93:913–20.
73. Harris L, Fritsche H, Mennel R, Norton L, Ravdin
P, Taube S, et al. American Society of ClinicalOncology 2007 update of recommendations forthe use of tumor markers in breast cancer. J ClinOncol 2007;25:5287–312.
74. Hough CD, Sherman-Baust CA, Pizer ES, Montz
FJ, Im DD, Rosenshein NB, et al. Large-scale serialanalysis of gene expression reveals genes differ-entially expressed in ovarian cancer. Cancer Res2000;60:6281–7.
75. Schummer M, Ng WV, Bumgarner RE, Nelson PS,
Schummer B, Bednarski DW, et al. Comparativehybridization of an array of 21,500 ovariancDNAs for the discovery of genes overexpressedin ovarian carcinomas. Gene 1999;238:375– 85.
76. Wang K, Gan L, Jeffery E, Gayle M, Gown AM,
Skelly M, et al. Monitoring gene expression pro-file changes in ovarian carcinomas using cDNAmicroarray. Gene 1999;229:101– 8.
77. Papsidero LD, Wang MC, Valenzuela LA, Murphy
GP, Chu TM. A prostate antigen in sera of pros-tatic cancer patients. Cancer Res 1980;40:2428 –32.
78. Kuriyama M, Wang MC, Papsidero LD, Killian CS,
Shimano T, Valenzuela L, et al. Quantitation ofprostate-specific antigen in serum by a sensitiveenzyme immunoassay. Cancer Res 1980;40:4658 – 62.
79. Potti A, Mukherjee S, Petersen R, Dressman HK,
Bild A, Koontz J, et al. A genomic strategy torefine prognosis in early-stage non-small-celllung cancer. N Engl J Med 2006;355:570 – 80.
80. Xu Y, Shen Z, Wiper DW, Wu M, Morton RE, Elson
P, et al. Lysophosphatidic acid as a potentialbiomarker for ovarian and other gynecologic can-cers. JAMA 1998;280:719 –23.
81. Villanueva J, Shaffer DR, Philip J, Chaparro CA,
Erdjument-Bromage H, Olshen AB, et al. Differ-ential exoprotease activities confer tumor-specific
serum peptidome patterns. J Clin Invest 2006;116:271– 84.
82. Leman ES, Cannon GW, Trock BJ, Sokoll LJ, Chan
DW, Mangold L, et al. EPCA-2: a highly specificserum marker for prostate cancer. Urology 2007;69:714 –20.
83. Mor G, Visintin I, Lai Y, Zhao H, Schwartz P,
Rutherford T, et al. Serum protein markers forearly detection of ovarian cancer. Proc Natl AcadS c iUSA 2005;102:7677– 82.
84. Petricoin EF, Ardekani AM, Hitt BA, Levine PJ,
Fusaro VA, Steinberg SM, et al. Use of proteomicpatterns in serum to identify ovarian cancer. Lan-cet 2002;359:572–7.
85. Esrig D, Elmajian D, Groshen S, Freeman JA, Stein
JP, Chen SC, et al. Accumulation of nuclear p53and tumor progression in bladder cancer. N EnglJ Med 1994;331:1259 – 64.
86. Malats N, Bustos A, Nascimento CM, Fernandez
F, Rivas M, Puente D, et al. P53 as a prognosticmarker for bladder cancer: a meta-analysis andreview. Lancet Oncol 2005;6:678 – 86.
87. Kim JH, Skates SJ, Uede T, Wong KK, Schorge JO,
Feltmate CM, et al. Osteopontin as a potentialdiagnostic biomarker for ovarian cancer. JAMA2002;287:1671–9.
Cancer Biomarkers from Bench to ClinicReviews
Clinical Chemistry 59:1 (2013) 157