evaluation of the novel serum markers b7-h4, spondin 2, and dcr3 for diagnosis and early detection...

06 (2007) 112–118www.elsevier.com/locate/ygyno

Gynecologic Oncology 1

Evaluation of the novel serum markers B7-H4, Spondin 2, and DcR3 fordiagnosis and early detection of ovarian cancer

Iris Simon a, Yan Liu b, Kirstin L. Krall a, Nicole Urban b, Robert L. Wolfert a,Nam W. Kim a,⁎, Martin W. McIntosh b

a diaDexus, Inc., 343 Oyster Point Blvd., South San Francisco, CA 94080, USAb Fred Hutchinson Cancer Research Center, Seattle, WA 98109, USA

Received 16 October 2006Available online 8 May 2007

Abstract

Objective. Early detection through regular screening could significantly reduce mortality from ovarian cancer. Advances in biomarkers andimaging continue to improve the sensitivity and specificity of cancer detection, but further improvements are still needed. In this study, weidentified and evaluated three new serum biomarkers that may be used to improve detection of ovarian cancer.

Methods. Through genomic analysis, we identified B7-H4, Spondin 2, and DcR3 as over-expressed genes in ovarian cancer tissues. Sensitivesandwich ELISAs were developed to analyze the level of these novel markers in 68 serum samples from patients with ovarian cancer (16 earlystage, 52 late stage) and 108 control samples, and 20 healthy women from which two serum samples were collected 1 year apart. CA125 levelswere measured in all samples.

Results. Markers were evaluated for their ability to identify clinical disease. The three novel markers and CA125 were elevated in serum ofovarian cancer patients as compared to normal controls. B7-H4 showed the highest specificity, with the lowest frequency of elevation in all controlgroups. When all cases were compared against all controls, CA125, Spondin 2, B7-H4, and DcR3 showed areas under the ROC curve of 0.87,0.78, 0.74, and 0.71, respectively. CA125 and B7-H4 showed the best diagnostic performance for early-stage, with AUCs of 0.90 and 0.80,respectively.

Conclusion. This study demonstrates that B7-H4, Spondin 2, and DcR3 are promising new ovarian cancer markers that may improve earlydetection of cancer when used in combination with traditional diagnostic tests.© 2007 Elsevier Inc. All rights reserved.

Keywords: Serum marker; Ovarian cancer; CA125 multivariate analysis; Tumor marker

Introduction

Ovarian cancer was diagnosed in 25,400 women and resultedin 14,300 deaths in 2004 in the US. Most cases present at anadvanced stage, and long-term survival is achieved in less than athird of patients [1]. However, the cure rate from currentlyavailable therapies is as high as 94% when the cancer is detected

Abbreviations: AUC, area under the curve; CV, coefficient of variation;ELISA, enzyme-linked immunosorbent assay; FPR, false positive rate; ROC,receiver operating characteristic; RT–PCR, reverse transcriptase PCR; TMB,tetramethylbenzidine; TVUS, transvaginal ultrasound⁎ Corresponding author. Fax: +1 650 246 6499.E-mail address: [email protected] (N.W. Kim).

0090-8258/$ - see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.ygyno.2007.03.007

in stage I [2]. Therefore, mortality could be significantlyreduced with detection at an early stage. The efficacy ofultrasound and the CA125 cancer marker combination todecrease mortality in screened populations is currently underinvestigation in several clinical trials [3,4]. Two studies foundthat among high-risk women, the sensitivity of transvaginalultrasound (TVUS) for detecting early-stage disease was only25% [5] and 31% [6], respectively. In addition, TVUS screeninghas generally been associated with a high rate of false positiveresults, leading to a large number of unnecessary surgicalprocedures [5–7]. However, the utility of CA125 as a screeningmarker is limited because it is frequently elevated in womenwith benign diseases such as endometriosis, liver disease andheart failure [8], and the use of CA125 and TVUS in

mailto:[email protected]

http://dx.doi.org/10.1016/j.ygyno.2007.03.007

113I. Simon et al. / Gynecologic Oncology 106 (2007) 112–118

combination showed that the predictive value of both methodscombined is relatively low [3,9,10]. Hence, there is a criticalneed for serum cancer markers that are more sensitive andspecific for the detection of ovarian cancer whether used aloneor in combination with CA125 and/or ultrasound [11].

Using cDNA database mining strategies that identifydifferentially expressed genes, and verification of the differen-tially expressed mRNAs by microarray and real-time quantita-tive RT–PCR, we identified novel genes that are over-expressedin ovarian cancer tissue, when compared with normal ovariantissues, or tissues from other organs [12]. We developedsensitive ELISA assays for three of these over-expressedproteins, DD-O110, DD-P108 and DD-C248 (known in theliterature as B7-H4, Spondin 2 and DcR3, respectively), tofurther evaluate their potential as ovarian cancer markers. B7-H4, a novel member of the B7-family, is expressed in activatedT-cells and may participate in the negative regulation of cell-mediated immunity in peripheral tissues [13–15]. Immunohis-tochemical studies showed membranous staining in serousovarian and breast cancer [16,17], and a direct role for B7-H4 inpromoting malignant transformation of epithelial cells wasdemonstrated [12]. B7-H4 was elevated in serum samples fromovarian cancer patients when compared to healthy controls orwomen with benign gynecological diseases. Moreover, inmultivariate logistic regression analyses, B7-H4 was additiveto CA125, particularly in the detection of early stage cancer[18]. Less is known about the role of Spondin 2 and DcR3 inovarian cancer. Spondin 2 is the human homologue of thezebrafish genes Mindin1 and Mindin2, which are members ofthe F-spondin superfamily of genes [19,20]. A recent publica-tion shows that Spondin 2 is expressed in normal prostate andover-expressed in prostate cancer, and suggests Spondin 2 aspotential target for antibody-based radiotherapy in prostatecancer [21]. In other studies, the Spondin 2 mouse homologuemindin was described as a pattern-recognition moleculeinvolved in the innate immune response to microbial pathogens[22]. In addition, new studies indicate that a mindin–integrininteraction is critical for inflammatory cell recruitment in vivo[23]. DcR3 (also known as TR6 and M68) is a secreted memberof the tumor necrosis factor receptor (TNFR) superfamily andmight inhibit FasL-mediated apoptosis [24]. It has been shownthat DcR3 is expressed in malignant cells of several tumor typesand its potential as a serum marker has been evaluated invarious cancer patient populations [25,26] but not in ovariancancer patients.

We have generated a series of monoclonal antibodies againstB7-H4, Spondin 2, and DcR3, and used them to develop a dualmonoclonal antibody sandwich ELISA for each marker. Thisstudy describes, for the first time, the assessment of Spondin 2and DcR3 as potential ovarian cancer markers, and confirmsfindings from earlier studies on the application of B7-H4 as aserum biomarker for ovarian cancer. These markers were testedon 176 serum samples to establish their elevation in ovariancancer patients when compared with healthy women or womenwith a benign condition. Their diagnostic performances werecompared to CA125 expression in univariate and multivariateanalyses.

Material and methods

Discovery of up-regulated genes in ovarian cancer

Proprietary bioinformatics algorithms were used to evaluate the LifeSeq(Incyte Corporation, Wilmington, DE) cDNA database for ESTs that werepresent preferentially in human ovarian cancer cDNA libraries compared to thelibraries from any other tissue types including normal ovary. Real-timeQuantitative RT–PCR (QPCR) reactions were performed using various tissuesamples to confirm and validate the markers with respect to cancer- and tissue-specific expression [12].

Generation of recombinant protein and specific monoclonalantibodies

The generation of recombinant proteins and specific monoclonal antibodiesagainst recombinant B7-H4 is described elsewhere [12]. Recombinant Spondin2 protein expressed in a baculovirus expression system, and recombinant His-tagged DcR3 protein expressed in mammalian cells were purified and used forfootpad immunization of BALB/c mice. Hybridoma supernatants were screenedby ELISA for reactivity against the target protein and for the absence of cross-reactivity to irrelevant proteins and family members. Pairing analysis studieswere performed to find specific antibodies for each protein that bind to twodifferent epitopes of the protein and could be used in sandwich ELISA formats.

Monoclonal antibody sandwich ELISA

The ELISA protocol for B7-H4 detection is described elsewhere [18], andthe ELISAs for the detection of Spondin 2 and DcR3 followed similar protocols.Briefly, 20–25 μL of undiluted serum sample was added to high bindingpolystyrene plates coated with capture mAb. Immobilized antigen was detectedwith biotinylated secondary mAb followed by horseradish peroxidase-conjugated streptavidin or alkaline phosphatase-conjugated streptavidin. Forcalibration, standards of recombinant protein and two controls were run inparallel with the test samples on each plate. CA125 levels were measured usingthe CA125 II assay from FujiRebio (FujiRebio Diagnostics, Inc., Malvern, PA)according to the manufacturer's protocol.

Human serum samples

All serum samples were collected by venipuncture technique from patientsand donors with appropriate informed consent. Sera from women with ovariancancer (n=68) and benign ovarian tumors (n=24) were collected before surgicalremoval of the ovaries, and stage and histology of the cancer were determined bystandard pathological methods [27]. Eighty-four controls without malignantovarian pathology came from healthy women who participated in an NCI-funded ovarian cancer screening research trial (n=41) and from women whounderwent surgery but were diagnosed with normal ovaries (n=39), corpusluteum (n=2) or functional cysts (n=2). These controls are combined in a sub-group termed ‘normal-ovary controls’. The ovarian cancer cases (n=68),randomly selected from a repository, included 16 patients diagnosed with earlystage cancer (stages I and II), while 52 (76.5%) cancers came from women withlate stage disease (stages III+IV).

Statistical analysis

Receiver operating characteristic (ROC) curve methods were used toquantify marker performance both graphically and statistically [27]. To aid in theinterpretation of our data when comparing markers, we transformed all markerswith the natural log so that expression in healthy subjects more accuratelyreflected a normal distribution. We then standardized the markers by assigning amean of 0 and unit standard deviation to the healthy controls [28].Standardization of the markers, which does not affect the ROC curves,facilitates the comparison of markers because of the uniformity of units ofmeasurement (the number of standard deviations above the average normalsubject).

114 I. Simon et al. / Gynecologic Oncology 106 (2007) 112–118

We defined composite markers (CM) as a weighted linear combination ofany two standardized markers, combining the novel markers with each other orwith CA125. We chose a linear combination for ease of interpretation. Logisticregression analysis was used to identify optimal marker combinations and toestimate the weights.

Logistic regression has several theoretical properties that make it convenientfor applied biomarker research [29], in particular its p-values, which help toevaluate whether the composite marker (CM) significantly increases the“distance” between cases and controls, as compared to the single marker. If themodel is correctly delineated, the sensitivity of the resulting CM is maximized atall specificities simultaneously, although the theoretically correct model cannotever be known in practice. After establishing significance, we examined theresulting ROC curves to evaluate the quality of the CM.

We also measured the temporal stability of a given marker in healthysubjects (n=20) by computing the Pearson correlation from two blood samplescollected 1 year apart. Temporal stability is an attribute of the marker itself thatgoes beyond assay stability; the latter, which is frequently affected by factorssuch as blood processing and storage conditions, is excellent in all of themarkers. Markers with high temporal stability, as measured by Pearsoncorrelation, can be expected to yield improved performance in a longitudinalalgorithm [30,31]. A high correlation, in particular one exceeding 0.5, impliesthat monitoring markers for their deviation from historical levels using theParametric Empirical Bayes (PEB) screening rule will yield earlier detectionthan a simple diagnostic rule that ignores screening history. However, onecannot conclude that one marker is better than another solely on its temporalstability.

Results

Single marker evaluation in controls

Of the total of 108 controls, 97 (90%) were post-menopausal,consistent with the menopausal status of the cancer group (88%post-menopausal). None of the three novel markers, B7-H4,Spondin 2, and DcR3, showed significant differences in pre-and post-menopausal groups.

For this study, three groups of controls were investigated tobest characterize the new markers in a population withoutovarian malignancy. As listed in Table 1, the smallest subgroupis patients with benign tumors of the ovary (n=24), who aremost likely to test positive on the markers. The largest subgroupis the surgical normal control group (n=43). These patients hadnormal ovarian pathology and were similar to cases with respectto conditions of the blood draw because their blood samples

Table 1Patient information

Normal-ovary controls Healthy Surgical normal a Total

41 43 84

Benign Serous Mucinous Brenner

10 3 1

Ovarian cancer Stage Serous Mucinous

I 3 2II 2 –III 23 1IV 5 –Unknown 1 –Total 34 3

a Patients with suspicious gynecological mass or bleeding, ovaries confirmed nor

were drawn just before surgery when they had experiencedsimilar stress, medications and an overnight hospital stay. Thefinal subgroup is healthy women participating in screening(n=41) in whom the markers are least likely to be elevated.Since statistical analyses suggested no difference in markerlevels between the control groups, screening and surgicalcontrols with normal ovaries were combined into a controlgroup termed “normal-ovary controls” (see Table 1).

In good agreement with results from larger studies [32],CA125 was below its cut-off point of 35 U/mL in all healthywomen but was elevated in some patients with benign ovariantumors and in women who underwent surgery. B7-H4 valueswere low in all non-ovarian cancer samples. The 95thpercentile for B7-H4 for normal-ovary controls and benigngroup was below 1 ng/mL, confirming results from studies withlarger groups of healthy women [18]. Levels of DcR3 andSpondin 2 were low in healthy women but showed elevation ina few women with benign ovarian diseases (Fig. 1 and listed inTable 2).

Single marker evaluation in cases versus controls

ROC curves for CA125, B7-H4, Spondin 2 and DcR3 aresummarized in Table 3 and are graphically displayed in Fig. 1E.CA125 had the greatest ability to distinguish cases fromcontrols. Its overall AUC was 0.87 and was increased to 0.90if only healthy controls were included in the control group.Spondin 2 showed the best diagnostic performance of the newmarkers when all stages of ovarian cancer samples were used,with an AUC of 0.84 when healthy normals were used ascontrols and 0.78 when all controls (normal-ovary controls plusbenign samples) were used. The ROC analysis for B7-H4resulted in an AUC of 0.74, and was not influenced by thecontrol group used for analysis, since B7-H4 had low values inall control groups. The mean and median value for B7-H4 inovarian cancer patients, even in patients with early stage disease,was higher than the 95th percentile of all control cases. DcR3showed significant separation between the cases and controlswith an AUC of 0.71 in the analysis of all cases versus allcontrols.

Non-neoplasm Other Total

3 7 24

Clear cell Endometrioid Undifferentiated Other Total

1 3 – 2 11– 2 – 1 51 1 5 8 39– 1 – 5 11– – – 1 22 7 5 17 68

mal through surgical examination.

Fig. 1. Distribution of serum CA125 (A), B7-H4 (B), Spondin 2 (C), and DcR3 (D) in healthy women (○), women with normal ovaries who underwent surgery (▵),women with benign ovarian tumors (+) and women with ovarian cancer (×). The distribution is shown using the standardized values for each marker. (E) ROC curvefor all markers using the differentiation of all controls from all ovarian cancer cases. The analysis resulted in AUCs of 0.87 (CA125), 0.78 (Spondin 2), 0.74 (B7-H4),and 0.71 (DcR3).


When only patients with early stage cancers (n=16) wereused for the ROC analysis, the ability of Spondin 2 and DcR3 todistinguish cases from controls decreased slightly, while thediagnostic performances of CA125 (AUC=0.9) and B7-H4(AUC=0.8) were equal or better than in all stages (Table 3).

Composite marker evaluation

The composite marker (CM) was defined as a weightedlinear combination of the standardized values of two markers,where logistic regression is used to estimate the weights. Linearregression p-values evaluate whether the CM compared to asingle marker significantly increases the distance between casesand controls [29]. All possible combinations of markers wereanalyzed and the significant combinations are listed in Table 4.Note that although combinations are determined to bestatistically significant, AUC improvements may over-estimatethe true improvement due to the statistical fitting, and will needto be confirmed in future studies. In this study, the combinationof three or more markers did not significantly improve theperformance of the CM over that of the two best markers (datanot shown).

Using only the 3 novel markers for analysis, the CM of B7-H4 and Spondin 2 showed significant improvement over thesingle markers. The composite marker, estimated by the logistic

regression, is defined as CM=0.45×(standardized B7-H4)+0.61×(standardized Spondin 2) and can be treated as a singlemarker. In ROC analysis, the AUC of the CM was 0.84, incomparison to AUC of 0.74 for B7-H4 and the AUC of 0.78 forSpondin 2 alone. Spondin 2 also improved the performance ofCA125 when used in combination (Table 4) by increasing theAUC to 0.90. In this analysis, CA125 carries 58% of the weightin the composite marker but the statistical analysis of the logisticregression shows that both markers are statistically significantand important contributors to the resulting CM ROC curve.

In the analysis of early stage cancers, the composite markerof CA125 and B7-H4 showed the best performance andexceeded the performance of each marker alone (Table 4).Using the specificity at 95% as a cut-off, CA125 was positive in11 women with early stage cancer (69%) and B7-H4 waselevated in 10 women (63%), while the composite markeridentified 13 cancers (81%). However, because of the lownumber of cases in this subgroup and the weight of B7-H4(32%), the p-value achieved (p=0.17) is not significant.

Longitudinal marker behavior

We evaluated the temporal stability of all individual markersin 20 women from whom serial serum samples were collected.The correlations are 0.72 (CA125), 0.36 (B7-H4), 0.61

Table 2Summary of raw data for CA125, B7-H4, Spondin 2 and DcR3

N Mean SD 95th percentile

CA125 (U/mL)Normal-ovary control 84 39.8 198.4 51.5Healthy 41 12.5 6.7 24.5Surgical normal 43 65.9 276.2 58.4

Benign 24 62.8 184.4 138.7Ovarian cancer 68 480.5 560.9 1426.3Serous 34 465.1 451.7 1413.2Non-serous 34 495.8 659.0 1780.1Early stage 16 516.2 736.1 2797.1Late stage 52 469.4 503.4 1426.3

B7-H4 (ng/mL)Normal-ovary control 84 0.42 0.68 0.76Healthy 41 0.38 0.19 0.69Surgical normal 43 0.47 0.94 0.76


Spondin 2 (ng/mL)Normal-ovary control 84 29.2 12.9 48.4Healthy 41 26.9 7.2 39.2Surgical normal 43 31.5 16.5 54.8

Benign 24 34.3 19.6 65.2Ovarian cancer 67 51.6 32 113.7Serous 34 51.2 37.6 121.9Non-serous 33 52 25.6 105.7Early stage 16 40.3 16.6 80.5Late stage 51 55.2 34.9 121.8

DcR3 (ng/mL)Normal-ovary control 84 1.22 3.58 1.61Healthy 41 1.62 5.1 1.61Surgical normal 43 0.84 0.59 1.50


Table 3Results from logistic regression analysis for single marker: AUC, sensitivity at95% and 98% specificity and Wilcoxon p-value for CA125, B7-H4, Spondin 2and DcR3 in all cancers and early stage cancers

Marker Control Cases AUC sen95 sen98 p_wilx

All cases versus healthy normalCA125 41 68 0.90 0.84 0.82 1.8e−10B7-H4 41 68 0.74 0.57 0.34 0.000083Spondin 2 41 67 0.84 0.54 0.43 4.8e−8DcR3 41 67 0.68 0.36 – a 0.0019

All cases versus normal-ovary controlsCA125 84 68 0.88 0.75 0.38 1.96e−13B7-H4 84 68 0.74 0.50 0.34 0.0000007Spondin 2 84 67 0.80 0.37 0.19 1.9e−9DcR3 84 67 0.70 0.36 0.04 0.00003

All cases versus all controls (normal-ovary controls+benign)CA125 108 68 0.87 0.74 0.38 3.3e−14B7-H4 108 68 0.74 0.40 0.34 0.0000002Spondin 2 108 67 0.78 0.28 0.19 2.2e−9DcR3 108 67 0.71 0.36 0.04 0.000007

Early stage cancer versus healthy normalCA125 41 16 0.93 0.87 0.87 0.000005B7-H4 41 16 0.79 0.69 0.44 0.002Spondin 2 41 16 0.76 0.50 0.31 0.004DcR3 41 16 0.59 0.31 –a 0.31

Early stage cancer versus all controls (normal-ovary controls+benign)CA125 108 16 0.90 0.69 0.37 0.000001B7-H4 108 16 0.80 0.63 0.44 0.0002Spondin 2 108 16 0.70 0.19 0.06 0.011DcR3 108 16 0.63 0.31 0.06 0.11

a Sensitivity at 98% specificity not available due to insufficient sample sizeand low control values.


(Spondin 2) and 0.89 (DcR3). All correlations are statisticallysignificant (p<0.05) except for B7-H4. The magnitude of thecorrelations for all other markers suggest that sensitivities maybe improved when used in screening algorithms that monitorchanges in baseline and the magnitude square-root (1-correla-tion) helps quantify the potential improvement; CA 125,Spondin 2, and DcR3 may be able to detect deviations whenthey are 52%, 61%, or 33% of the size detected when using onlysimple single-threshold screening rules [27].

Discussion

We evaluated the novel ovarian cancer markers DD-O110,DD-P108 and DD-C248 (also known in the literature as B7-H4,Spondin 2 and DcR3, respectively) for several performancecharacteristics relevant to their potential use in diagnosis and

early detection of ovarian cancer. All three markers showedelevation in serum of ovarian cancer patients when compared tohealthy women, consistent with the predictions made fromanalysis of expression at the transcriptional level. Analysis ofB7-H4 performance in this study confirmed our earlier resultsthat B7-H4 is a promising novel ovarian cancer marker withminimal elevation in healthy women or women with benignovarian tumors [18]. The best individual marker for diagnosiswas found to be CA125, which performed marginally better inthis study than in previously published analyses. Of the novelmarkers, B7-H4 and Spondin 2 showed the highest level ofdiscrimination of cancer cases. B7-H4 had higher sensitivities at95% specificity and performed better than Spondin 2 in thedetection of early stage cancer. DcR3 also showed significantelevation in ovarian cancer compared to the normal controls. Inthe analysis of all cancers and controls, the composite markercombining Spondin 2 and CA125 was superior to CA125 alone.The CM had higher sensitivity at high specificity (improvingsensitivity from 38% for CA125 alone to 55% for the CM at98% specificity).

In the detection of early stage cancer, the composite markercombining B7-H4 and CA125 showed the best performance andhad higher sensitivity than CA125 alone. Since most of thecases in this study were diagnosed as stage III or IV, larger

Table 4Results from linear regression analysis of composite marker (CM), AUC,sensitivity at 95% and 98% specificity, p-value (p1=p-value for first marker,p2=p-value for second marker) and weight of single marker using thestandardized (sd) values

Marker Controls Cases AUC Sens95

Sens98

P1 P2

All cancer versus all controls (normal-ovary controls+benign)B7-H4+Spondin 2 108 67 0.84 0.54 0.34 0.0001 <0.0001CA125+Spondin 2 108 67 0.90 0.72 0.55 <0.0001 0.0091

Early stage cancer versus all controls (normal-ovary controls+benign)CA125+B7-H4 108 16 0.91 0.81 0.56 <0.0001 0.17

CM=0.45*sd B7-H4+0.61*sd Spondin 2.Weight for B7-H4=0.42, weight for Spondin 2=0.58.CM=0.55*sd CA125+0.40*sd Spondin 2.Weight for CA125=0.58, Weight for Spondin 2=0.42.CM=0.57*sd CA125+0.27*sd B7-H4.Weight for CA125=0.68, weight for B7-H4=0.32.


studies are needed to confirm this conclusion. The AUC forCA125 in all cases versus all controls was 0.87 with 74%sensitivity at 95% specificity, which is in good agreement withpublished studies [32]. However, when early stage cancers werecompared to all controls, CA125 showed a sensitivity of 69% at95% specificity. Nearly all stage I patients had positive CA125values and, therefore, exceeded the 50% positive detection ratenormally observed in this group. Because CA125 detected anunusually high proportion of stage I cancers in this study, it wasdifficult to accurately determine the additive values of othermarkers in the detection of early stage cancers, or in the overallpatient group.

Temporal stability data showed that the correlations of allmarkers were statistically significant except for B7-H4,suggesting that the level of B7-H4 in individuals may be morevariable over time than that of the other markers examined in thisstudy. However, a previous study on 17 healthy pre-menopausalwomen, from which 7–9 blood draws per person were collectedover a period of 1month, showed that B7-H4 had relatively goodbiological stability with intra-individual (variation within asubject) CVs between draws ranging from 0.3% to 7.9%(average CV=2.9%) [33]. In addition, serum samples taken3 years apart from 11 healthy subjects showed intra-individualCVs ranging from 0.1% to 40% (average CV=14%), againshowing good long-term stability (data not shown). It should benoted that these data were generated with healthy pre-menopausal women, and the biological variation and long-term temporal stability in post-menopausal women may bedifferent. Thus, temporal stability of B7-H4 and other markers inpost-menopausal women should be studied in more detail infuture studies to determine significant relevance to ovariancancer detection.

Retrospective studies have shown that women with ovariancancer are more likely than controls to have unspecificsymptoms such as abdominal swelling and pain for up to6 months before diagnosis [34]. The evaluation of symptomaticwomen with a specific cancer marker may therefore simplify thetreatment of these patients and increase the detection rate ofcancers at an earlier stage. The novel serummarkers described in

this study should be evaluated in high-risk symptomatic womento facilitate the detection of ovarian cancer. Ultimately, a pro-spective study on asymptomatic women is required to determinewhether any marker, or panel of markers, will be able to identifywomen with ovarian cancer at the early stages when the cancer ismost treatable [10]. Promising results from such an approachwas obtained in a clinical study to address early elevation ofCA125 in asymptomatic women [35]. Such longitudinal studieswill be a logical next step for evaluation of novel markers such asthe ones that were described in this study. Because samples fromlongitudinal studies are very valuable, all new markers have tobe rigorously evaluated in several independent studies beforethey can be tested in these samples [36].

In this study, we have evaluated three novel serumbiomarkers for ovarian cancer in a well-characterized sampleset. The results indicate that all three markers are elevated inovarian cancer when compared to normal and benign controls.These results support the application of the novel markers asmembers of a multi-marker panel to improve detection ofovarian cancer. Results from these and other studies may lead toa sensitive and accurate blood test for early detection of ovariancancer that could have significant impact by increasing thesurvival rate of patients suffering from this devastating disease.

Acknowledgments

We thank the Protein Science and Antibody Department atdiaDexus for the generation of key reagents and we thank IrenaKing for technical assistance in analyzing serum samples forCA125II levels. The serum samples for our analysis wereprovided by the Pacific Ovarian Cancer Research Consortium, aSpecialized Program of Research Excellence (SPORE) inovarian cancer (NCI P50 CA83636, PI: Nicole Urban).

References

[1] Edwards BK, Brown ML, Wingo PA, Howe HL, Ward E, Ries LAG, et al.Annual report to the nation on the status of cancer, 1975–2002, featuringpopulation-based trends in cancer treatment. J Natl Cancer Inst2005;97:1407–27.

[2] American Cancer Society. Cancer facts and figures 2005 [online]. Atlanta,GA, available from: URL: http://www.cancer.org/docroot/stt/stt_0.asp.

[3] Buys SS, Partridge E, Greene MH, Porok PC, Reding D, Riley TL, et al.Ovarian cancer screening in the Prostate, Lung, Colorectal and Ovarian(PLCO) cancer screening trial: findings from the initial screen ofrandomized trial. AJOG 2005;193:1630–9.

[4] Jacobs IJ, Menon U. Progress and challenges in screening for earlydetection of ovarian cancer. Mol Cell Proteomics 2004;3.4:355–66.

[5] Bell R, Petticrew M, Sheldon T. The performance of screening tests forovarian cancer: results of a systematic review. Br J Obstet Gynaecol1998;105:1136–47.

[6] van Nagell Jr JR, DePriest PD, Reedy MB, Gallion HH, Ueland FR, PavlicEJ, et al. The efficacy of transvaginal sonographic screening inasymptomatic women at risk for ovarian cancer. Gynecol Oncol2000;77:350–6.

[7] Sato S, Yokoyama Y, Sakamoto T, Futagami M, Saito Y. Usefulness ofmass screening for ovarian carcinoma using transvaginal ultrasonography.Cancer 2000;89:582–8.

[8] Markman M. The role of CA 125 in the management of ovarian cancer.Oncologist 1997;2:6–9.


[9] Stirling DD, Evans GR, Pichert G, Shenton A, Kirk EN, Rimmer S, et al.Screening for familial ovarian cancer: failure of current protocols todetect ovarian cancer at an early stage according to the InternationalFederation of Gynecology and Obstetrics system. J Clin Oncol 2005;23:5588–5596.

[10] Jacobs IJ. Screening for familial ovarian cancer: the need for well-designedprospective studies. J Clin Oncol 2005;24:5443–5.

[11] Bast RC, Urban N, Shridhar V, et al. Early detection of ovarian cancer:promise and reality. In: Stack MS, Fishman DA, editors. Ovarian cancer.Cancer research and treatment, vol. 107. Boston: Kluwer AcademicPublishers; 2001. p. 61–97.

[12] Salceda S, Tang T, Kmet M, et al. The immuno-modulatory protein B7-H4is overexpressed in breast and ovarian cancers and promotes epithelial celltransformation. Exp Cell Res 2005;306:128–41.

[13] Zang X, Loke P, Kim J, Murphy K, Waitz R, Allison JP. B7x: a widelyexpressed B7 family member that inhibits T cell activation. Proc Natl AcadSci 2003;100:10388–92.

[14] Sica GL, Choi IH, Zhu G, Tamada K, Wang SD, Chapoval AI, et al. B7-H4, a molecule of the B7 family, negatively regulates T cell immunity.Immunity 2003;18:849–61.

[15] Collins M, Ling V, Carreno BM. The B7 family of immune-regulatoryligands. Genome Biol 2005;6:223.1–7.

[16] Tringler B, Zhuo S, Pilkington G, et al. B7-H4 is highly expressed in ductaland lobular breast cancer. Clin Cancer Res 2005;11:1842–8.

[17] Tringler B, Liu W, Corral L, et al. B7-H4 over-expression in ovariancancer. Gynecol Oncol 2006;100:44–52.

[18] Simon I, Zhou S, Corral L, Diamandis EP, SarnoMJ, Wolfert RL, et al. B7-H4 is a novel membrane-bound protein and a candidate serum and tissuebiomarker for ovarian cancer. Cancer Res 2006;66:1570–5.

[19] Feinstein Y, Borrell V, Garcia C, Burstyn-Cohen T, Tzarfaty V, Frumkin A,et al. F-spondin and mindin: two structurally and functionally related genesexpressed in the hippocampus that promote outgrowth of embryonichippocampal neurons. Development 1999;126:3637–48.

[20] Klar A, Baldassare M, Jessell TM. F-spondin: a gene expressed at highlevels in the floor plate encodes a secreted protein that promotes neural celladhesion and neurite extension. Cell 1992;69:95–110.

[21] Parry R, Schneider D, Hudson D, et al. Identification of a novel prostatetumor target, mindin/RG-1, for antibody-based radiotherapy of prostatecancer. Cancer Res 2005;65:8397–405.

[22] He YW, Li H, Zhang J, et al. The extracellular matrix protein mindin is a

pattern-recognition molecule for microbial pathogens. Nat Immunol2004;5:88–9.

[23] Jia W, Li H, He YW. The extracellular matrix protein mindin serves as anintegrin ligand and is critical for inflammatory cell recruitment. Blood2005;106:3854–9.

[24] Pitti RM, Marsters SA, Lawerence DA, Roy M, Kischkel FC, Dowd P.Genomic amplification of a decoy receptor for Fas ligand in lung and coloncancer. Nature 1998;396:699–703.

[25] Wu Y, Han B, Sheng H, Moore PA, Zhang J, Wu J. Clinical significance ofdetecting elevated serum DcR3/TR6/M68 in malignant tumor patients. IntJ Cancer 2003;105:724–32.

[26] Chen J, Zhang L, Kim S. Quantification and detection of DcR3, a decoyreceptor in TNFR family. J Immunol Methods 2004;285:63–70.

[27] McIntosh MW, Drescher C, Karlan B, Scholler N, Urban N, Hellstrom KE,et al. Combining CA125 and SMR serum markers for diagnosis and earlydetection of ovarian cancer. Gynecol Oncol 2004;95:9–15.

[28] McIntosh MW, Pepe M. Combining several screen tests: optimality of therisk score. Biometrics 2002;58:657–64.

[29] Sato A, McIntosh M. Evaluating the performance of adaptive compared tonon-adaptive screening rules for cancer when using serum biomarkers.University of Washington: Department of Biostatistics Seattle; 2004.

[30] Skates CJ, Pauler DK, Jacobs I. Screening based on the risk of cancercalculation from Bayesian Hierarchical Change-point models of long-itudinal markers. J Am Stat Assoc 2001;96:429–39.

[31] McIntosh M, Urban N, Karlan B. Generating longitudinal screeningalgorithms using novel biomarkers for disease. Cancer Epidemiol,Biomarkers Prev 2002;11:159–66.

[32] Jacobs I, Bast Jr RC. The CA125 tumour-associated antigen: a review ofthe literature. Hum Reprod 1989;4:1–12.

[33] Simon I, Kim NW, Kuo T, et al. Ovr110: a potential serum marker for earlydetection of ovarian cancer. AACR 95th Annual Meeting 2004 Proceed-ings, vol. 45; 2004. p. 913–4. [abstr.].

[34] Smith LH, Morris CR, Yasmeen S, Parikh-Patel A, Cress RD, Romano PS.Ovarian cancer: can we make the clinical diagnosis earlier? Cancer2005;104:1398–407.

[35] Jacobs IJ, Skates SJ,MacDonaldN,MenonU, Rosenthal U, Davies A, et al.Screening for ovarian cancer: a pilot randomized controlled trial. Lancet1999;353:1207–10.

[36] Urban N, Drescher C. Current and future developments in screening forovarian cancer. Women's Health 2006;2(5):733–42.

evaluation of the novel serum markers b7-h4, spondin 2, and dcr3 for diagnosis and early detection...

Documents