green journal roma

8/4/2019 Green Journal ROMA

1/9

Evaluation of the Diagnostic Accuracy of theRisk of Ovarian Malignancy Algorithm inWomen With a Pelvic MassRichard G. Moore, MD, M Craig Miller, BSc, Paul Disilvestro, MD, Lisa M LandrumJMD,Walter Gajewski, MD,]ohnj Ball, MD, and Steven] Skates, Plt.DOBJECTIVE: It is often difficult to distinguish a benignpelvic mass from a malignancy and tools to help referringphysician are needed. The purpose of this study was tovalidate the Risk of Ovarian Malignancy Algorithm inwomen presenting wi th a pelvic mass.METHODS: This was a prospective, multicenter, blindedclinical trial tha t included women who presented to agynecologist. a family practitioner, an internist, or ageneral surgeon with an adnexal mass. Serum HE4 andCA 125 were determined preoperatively. A Risk of Ovarian Malignancy Algorithm score was calculated and classified patients into high-risk and low-risk groups forhaving a malignancy. The sensitivity, specificity, negativepredictive value, and positive predictive value of the Riskof Ovarian Malignancy Algorithm were estimated.RESULTS: A total of 472 patients were evaluated with 383women diagnosed with benign disease and 89 womenwith a malignancy. The incidence of all cancers was 15%and 10% for ovarian cancer. In the postmenopausalFrom the Department of Obstetria and Gynecology and t/ze Cmter forBio11U1rlws and Emergzng Tedmologies, Program in Wo= s Oncology, Womenand hifants' /fospita/1 Brown University, Providence, Rlwtk Island; the Sectionof Gynecology Oncvlogy, Dcpartme11l ofObstetrics and G y ~ U e o l o g y , OkkhomaUni!JtT$ity Hea/Jh Stitnet Center, 0/dalwma GUy, 0/dalwma; Zimmer CancerCtnl.er, New Honorer &giolllll Medical Center, Wilmington, North Carolina;jackson Clime, jadc.son, T t J ~ J ~ U K t ; and the Cancer Center, MassadmseUsGeneral HospWJJ., HarMTd Medical Sdwol, Boston, Massacltusetts.Funded by Fujirtbio D1agnostiainc. RiduudG. Moore ispartiaU, supportedbyNational Concer Institute (NCI) grant CA73649HJ1 and Steven}. Skatts ISpartiaU, supported by NCJ grant CA T 2990.Correspondingauthor: Richard G. Moore, MD,Program in Womt11:S Onmlogy,Womm and lnfa11ts' Hosplla/, Brown Universily, PrOIJidmce, R1 02925;e-maiL [email protected]. Moqre receives research funding from Fujirtbio Diagnostia Inc and AbbottDiagnostics Inc. M. Craig Miller receives consulting fees from FujirthioDiagnostics Inc. Dr .Boll received research funding from Fujirthio Diagnostioinc. TM other au/Mrs did not report atiJ potential conflicts of nterest.C 2017 by flte Ammcan College ofObsUtriciansand Gynecologists. Publishedby LippitiCQtt Williams & Wil/..1"ns.ISSN: 0029-7844/11

280 VOL 118, NO. 2, PART 1 AUGUST 2011

group, a sensilivity of 92.3% and a specificity of 76.0%and for the premenopausal group the Risk of OvarianMalignancy Algorithm had a sensitivity of 100% andspecificity of 74.2% for detecting ovarian cancer. Whenconsidering all women together, the Risk of OvarianMalignancy Algorithm had a sensitivity of 93.8%, a specificity of 74.9%, and a negative predictive value of 99.0%.CONCLUSION: The use of the serum biomarkers HE4and CA 125 with the Risk of Ovarian Malignancy Algorithm has a high sensitivity for the prediction of ovariancancer in women with a pelvic mass. These findingssupport the use of the Risk of Ovarian MalignancyAlgorithm as a tool for the triage of women with anadnexal mass to gynecologic oncologists.(Obstet Gynecol 2011;118:280--8)DOl: 10.10971AOG.Ob013e318224fce2LEVEL OF EVIDENCE: II

0 varian cancer is lhe leading cause of death fromgynecologic malignancies in the United Stateswith an annual incidence of 22,000 cases an d anannual mortality rate of approximately 14,000.1 Theoptimal treatment for ovarian cancer includes cytoreductive surgery followed by adjuvant chemotherapywith significantly belter prognostic outcomes i f theinitial surgery and treatment are performed by surgeons and at centers experienced in the managementof ovarian cancer.2-6

At the time of the initial diagnosis, women withovarian cancer symptoms and adnexal masses presentprimarily to gynecologists, primary care physicians,or general surgeons. Triage guidelines put forth by theAmerican College of Obstetricians an d Gynecologistsand the Society of Gynecologic Oncologists recommend referral of women with a pelvic mass at highrisk for ovarian cancer to a gynecologic oncologist.78An important dilemma is faced by the physicians wh oinitially sec these patients as to which patients are

OBSTETRICS & GYNECOLOGY


2/9

appropriate to refer to a gynecologic oncologist Thebiage of women at high risk for an ovarian malignancy is of vital importance because recent studiesindicate that patients with ovarian cancer managed bygynecologic oncologists and at high-volume institutions are more likely to undergo complete surgicalstaging and optimal cytoreductive surgery with fewercomplications and better survival rates than patientstreated by surgeons less familiar with the management of ovarian cancer.236.9-t 2 To improve the triageof patients with pelvic masses, we performed a pilotstudy that. measured seven biomarkers in sera drawnfrom women with a pelvic mass before surgery. Wefound that when adding an additional biomarker toCA 125, only HE4 was able to increase the sensitivitycompared with using CA 125 alone. 13 Data from thispilot study and a second pilot study were combined todevelop the Risk of Ovarian Malignancy Algorithm,which combines serum levels of the biomarkers CA125 and HE 4 along with menopausal status in alogistic regTession model to classify patients with apelvic mass into high-risk or low-risk groups forhaving ep1thelial ovarian cancer. The clinical performance of the Risk of Ovarian Malignancy Algorithmwas first validated in a prospective, double-blindclinical study on patients with an adnexal mass re cruited at tertiary care centers in patients who presented to gynecologic oncologists and thus wereconsidered a high-risk cohort. 14 At a set specificity of75%, the Risk of Ovarian Malignancy Algorithmdisplayed a sensitivity of 94% for distinguishingbenign status from epithelial ovarian cancer and85% sensilivity lo identify early stage I and TIdisease. Additionally, the Ri sk of Ovarian Malignancy Algorithm outperformed the Risk of Malignancy Index, an algorithm that uses ultrasound,menopausal status, and CA 125 for differentiatingbenign from malignant disease. 15

The purpose of the cuiTent study was to validatethe use of the Risk of Ovarian Malignancy Algorithmin a low-risk population of women presenting to ageneralist with an adnexal mass and evaluating theclinical use of the Risk of Ovarian Malignancy Algorithm in aiding the triage of women to gynecologiconcologists.MATERIALS AND METHODSThis study was a prospective, multicenter, doubleblind clinical bial conducted at 13 clinical study sites(seven general and six specialty hospitals) across theUniled StaLes between October 2009 and August2010. The study protocol was approved by institutional review boards at each site and was registered on

the Clinica!Trials.gov web site (accession numberNCT00987649). Written informed consent was obtained from each patient before entry into the trialand the collection of blood.

The study included premenopausal and postmenopausal women 18 years of age or older presenting to a generalist (defined as a general gynecologist,internist, fn.miJy practitioner gastroenterologist, orgeneral surgeon) with an ovarian cyst or an adnexalmass and subsequently scheduled to undergo surgery.An adnexal mass was defined as a simple, complex,or a solid ovarian cyst or any mass in the pelvis asdetermined by imaging {ie, ultrasonography, computed tomography scan. or magnetic resonance imaging). Menopausal status was determined by thephysicians through history and physical examination.l f the menopausal status was no t known or reported,the following criteria were used: women 48 years oryounger were considered premenopausal and women55 years and older were considered postmenopausal.I f the patient's age was between 49 and 55 years andthe last menstrual cycle was unknown, the folliclestimulating hormone levels were analyzed. Women\.vith follicle-stimulating hormone levels greater than22 milli-international units/mL were considered postmenopausal and lower than 22 were considered premenopausal as recommended in the U.S . Food andDrug Administration package insert for the AbbottARCIDTECT i2000 platform. Women with a historyof ovarian cancer, bilateral oophorectomy, currentlyknown to be pregnant, or unable to provide informedconsent were excluded from the study.

Blood samples were collected from all pal.ientsswithin 30 days before their surgical procedures, andall samples were drawn before the induction of anesthesia. Blood was collected into Serum SeparatorTubes and was allowed to clot for ar least 30 minutesbefore centrifugation. TI1e blood samples were centrifuged for 10 minutes at l,l00 - 1,300 g, serum andplasma were separated, transferred to cryovials, andfrozen at 20C until testing. Appropriate volumes ofserum were tested to quantitate lhe concentrations ofCA 125 and HE4. Serum HE4 levels were determined using an enzyme-linked immunoabsorbent assay kit and serum CA 125 levels were determinedusing an ARCHITECT CA125II assays on the ARCHITECT Instrument All the samples were tested induplicates. Coefficient of variation for all the tests didnot. exceed 15.0%.

The primary end point of the current clinicalstudy was lo determine the effectiveness of the rusk ofO varian Malignancy Algorithm for predicting epithelial ovarian cancer in patients presenting to a gener-

VOL. 118, NO. 2, PART 1, AUGUST 201 1 Moore et al Use of he Risk of Ovarian Malignancy Algorithm 281


3/9

alist with a pelvic mass. A detailed description of thedevelopment of the Risk of Ovarian MalignancyAlgorithm and statistical approach has been described in a prior publication by Moore et al.H

To calculate the Risk of Ovarian MalignancyAlgorithm score, a predictive index was calculatedusing Lhe serum HE4 and serum CA 125II levels an done of the following equations, depending on thepatient's menopausal status:

1. Premenopausal: predictive index (PI)= -12.0+2.38*LN[HE4]+0.0626*LN[CA 125)

2. Postmenopausal: predictive index (PI)=-8.09+l.04*LN[HE4]+0.732*LN[CA 1251Th e following equation used the predictive index foreach patient to calculate a Risk of Ovarian Malignancy score:

ROMA score (%)= exp(PI)/[1+exp(PI)]* 100Th e Risk of Ovarian Malignancy Algorithm was

used to stratify women into high-risk or low-riskgroups for having a pelvic mass that is malignant orbenign respectively. For both statistical and medicalreasons, cut points were chosen that provided a setspecificity of750fo for the .HE4 EIA and ARCHITECTCA 125 II assay panel. 14 In the original pilot bial andprior validation trial, the Risk of Ovarian MalignancyAlgorithm score thresholds of 13.1% or greater and27.7% for premenopausal and postmenopausalwomen, respectively, achieved a se t specificity of 75%as d e s i r e d . Thus, for premenopausal patients, aRisk of Ovarian Malignancy Algorithm score of13.1% or greater is considered high risk for malignancy and for postmenopausal patients, a Risk ofOvadan Malignancy Algorithm score of 27.7% orgreater is considered high risk for malignancy.

Preoperative Risk of Ovarian Malignancy Algorithm scores were calculated and the sensitivity, specificity, positive predictive value, and negative predictive value were determined for all groups andsubgroups. Th e Wilcoxon rank-sum test was used forthe comparison of benign compared with malignantgroups and between menopausal status when examining serum levels ofHE4 and CA 125. A Pvalue of< .05 was considered as statistically significant. The Pvalues were not adjusted for multiple evaluations.RESULTSThirteen diverse sites from across the United Statedenrolled 512 women with a pelvic mass of which 472(92.2%) were evaluable and are the focus of thisreport. Th e demographics for this cohort and pathologic classification are presented in Tables 1 2,respectively. There were 255 premenopausal patients

and 217 postmenopausal patients. Plasma follicleslimulaling hormone levels were used to determinemenopausal status in 36 women, 22 of which had atleast a remaining ovary after a prior hysterectomy.The mean age of all women entered on the trial was50.3 years (range 18- 89 years). The mean age for thesubgroup of premenopausal women was 39.7 years(range 18 - 56 years) and for the subgroup of postmenopausal women 62.8 years (range 44-89 years).Women diagnosed with benign disease made up81.1 Ofo (383) of lhc cohort (150 postmenopausal and233 premenopausal), and women diagnosed with amali&rnancy or a low malignant potential tumor madeup 18.9% (89) of the cohort (67 postmenopausal an d22 premenopausal). Th e histologic classification ofthe benign pathology is provided in Table 1 andmalignant disease along with the primary site isdisplayed in Table 2.

Forty-eight women {39 postmenopausal and ninepremenopausal) were diagnosed with a primary ovarian, fallopian tube, or primary peritoneal cancer allcategorized under epithelial ovarian cancer. The stageof these cancers was as follows: eight stage I, fourstage II , 32 stage III, two stage N, and two unstaged.Low malignant potential tumors were diagnosed in 19patients (12 postmenopausal and seven premenopausal), two patients were diagnosed with nonepithclial ovarian cancer, 11 patients with other gynecologiccancers, and nine patients with nongynecologic cancers (Table 2).

Evaluation of serum levels of HE4 and CA 125was carried ou t for all patients and for separatemenopausal groups (Table 3). Significant differencesfor serum HE 4 and CA 125 levels were detectedwhen comparing benign cases with all ma lignantcases or wilh all epithelial ovarian cancers and lowmalignant potential tumors together {all .?


4/9

Table 1. Patie nt Demographics an d Benign Tumor Characterist icsPremenopausal

RaceWhite 202African American 23Hispanic 8Asian 12, ative American 3Other 5Unknown 2Total 255Benign diseaseAdenofibroma 1Adenomyosis 1Benign cyst 1Corpus luteum 26Cystadenofibroma 0Cystadenoma 0Endometriosis 54leiomyoma 11ribroma 5Fibrothecoma 0Follicular cyst 9Hemorrhagic cyst 7Inclusion cyst 2Mature teratoma 23Mucinous cystadenoma 13ormal ovaries 4Paratubal cyst 9Serous cystadenoma 33Simple cyst 23STUMP 1TOA 6Torsion 0Other 4Total 233

Data are n or n (%).STUMP, smooth muscle tumor of uncertain ma lignant potential.TOA. tubo-ovarian abscess.

the low-risk group providing a sensitivity of 88.1%(95% confidence interval [CI] 77.8-94.7%), a specificity of 74.9% (95% CI 70.3-79.2%), and a negativepredictive value of 97.3% (95% CI 94.7-98.8%) asillustrated in Table 5. The algorithm incorrectly classified five of 19 patients with low malignant potentialtumors and only three of 48 wil.h epithelial ovariancancer (all three stage I-II) to the low-risk group,providing a sensitivity for epithelial ovarian cancer of93.8% (95% CI 82.8-98.7%) at a specificity of 74.9%(95% CI 70.3-79.2%) and a negative predictive valueof 99.00/o (95% CI 97.0-99.8%) (Table 6).

Examination of postmenopausal women with benign disease (n= 150) or with epithelial ovarian cancerand low malignant potential tumors (n=51} revealedthat the Risk of Ovarian Malignancy Algorithm classified 46 of the 51 women with epithelial ovarian

Postm enopausal Al l Palients

197 399 (84.5)9 32 (6.8)7 15 (3.2)1 13 (2.8l2 5 (1.1)1 6 (1.2)0 2 (0.4)

217 472 (100.0)0 1 (0.3)1 2 (0.5)1 2 (0.5)2 28 (7.3)1 1 (0.3)2 2 (0.5)13 67 (17.5)5 16 (4.2)6 11 (2.9)3 3 (0.8)0 9 (2.3)1 8 {2.1)1 3 (0.8)7 30 (7.8)

17 30 (7.8)7 11 (2.9)4 l3 (3.4)

62 95 (24.8)8 31 (8.1)0 1 (0.3)8 14 (3.7)1 1 (0.3)0 4 (1.0)

150 383 (100.0)

cancer or low malignant potential tumors into thehigh-risk group and 114 of the 150 women withbenign tumors into the low-risk group providing asensitivity of 90.2% (95% CI 78.6-96. 7%}, a specificityof 76.0% (95% CI 68.4-82.6%), and a negative predictive value of 95.8% (95% CI 90.5-98.6%) as illustrated in Table 5. Th e algorithm incorrectly classifiedtwo of 12 patients with low malignant potential tu-mors and only three of 39 wil.h epithelial ovariancancer {all three stage I-ll) to the low-risk group,providing a sensitivity for epithelial ovarian cancer of92.3% (95% CI 79.1-98.4%) at a specificity of 76.0%(95% CI 68.4-82.6%) and a negative predictive valueof 97.4% (95% CI 92.7-99.5%) (Table 6).Examination of premenopausal women with benign disease (n=233) or with epithelial ovarian cancerand low malignant potential tumors {n= l6) revealed

VOL. 118, NO . 2, PART 1, AUGUST 2011 Mo ore et al Use of the Risk of Ovarian Malignancy Algorithm 283


5/9

Table 2. Malignant Tumo r Characte risticsPremenopausal Postmenopausal All Patients

CancerBorderline or low malignant potential tumorE p i t h e l i < ~ l ovarian cancerNonepithelial ovarian cancerOther gynecologic cancerOther cancerMetastatic cancerTotalEOC grade123TotalEpithelia l ovarian cancer stageIIIIl lIVUnstagedTotal

EOC, epithelial ovarian cancer.

that the Risk of Ovarian Malignancy Algorithm classified 13 of the 16 women with epithelial ovariancancer or low malignant potential into the high-riskgroup and 173 of the 233 women with benign tumorsinto the low-risk group providing a sensitivity of81.3% (95% CI 54.4-96.0%), a specificity of 74.2%(95% Cl 68.1-79.7%), and an negative predictivevalue of 98.3% (95% Cl 95.1-99.6%) as illustrated inTable 5. Th e algorithm incorrectly classified three ofseven patients with low malignant potential tumors.Al l patients with epithelial ovarian cancer were correclly classified into the high-risk group. Thus, of thenine premenopausal patients with epithelial ovariancancer, aU were correctly classified providing a sensitivity for epithelial ovarian cancer of lOOOfo (95% CI66.4-100%) at a specificity of 74.2% (95% CI 68.1-Table 3. Medians and Ranges fo r HE4 and CA125

7 12 19 (2 1.3)9 39 48 (53.9)0 2 2 (2 .2 )3 8 11 (12.4)1 6 7 (7 .9)2 0 2 (2.2)22 67 89 (100.0)5 5 10 (2 0.8)0 9 9 (18.8)4 25 29 (60.4)9 39 48 (100.0)2 6 8 (16.7)1 3 4 (8.3)5 27 32 (66.6)0 2 2 (4.2)I 1 2 (4.2)9 39 48 (100.0)

79.7%) and a negative predictive value of 100% (95%CI 97.9-100%) (Table 6).

On consideration of women with early-stage disease alone (stage I and D), the Risk of OvarianMalignancy Algorithm classified nine of 12 patientswith epithelial ovarian cancer correctly into the highrisk group, thus achieving a sensitivity for early stageepithelial ovarian cancer of 75.0% (95% CI 42.8-94.5%) , a specificity of 74.9% (95% Cl 70.3-79.2%) ,and a negative predictive value of 99.0% (95% Cl97.0-99.8%) . When including early-stage epithelialovarian cancer and low malignant potential tumors,the Risk of Ovarian Malignancy Algorithm classified23 of 31 patients wilh epilhelial ovarian cancer or lowmalignant potential tumors colTectl) lo the high-riskgroup, thus achieving a sensitivity of 74.2% (95% CI

Median (Range)HE4 (pM) CA 125 (International Un its/ml )

All benignPremenopausalPostmenopausal58.1 (27 .1-403.2)53.8 (27.8-118.2}71.2 (32.8-403.2)

19.9 (3.6-1,085 .1 )22.2 (3.6-1,085 .1 )17.2 4 . 3 ~ 8 9 . 1366.8 (1 1.9- 1,073.9)All epithelial ovarian cancer and low malignantpotential tumorsPremenopausalPostmenopausalAll cancersPremenopausalPostmenopausal

266.8 (31.2-13,250.0)252.2 (31.3-4,632.0)266.8 (32.8-1,776.7)242.5 (43.1 13,250.0)148.7 (47. 2-400.8)266 .8 (43.1-13,250.0)

284 Moo re e l al Use of the /?isk of Ovarian Malignancy Algorithm

449.1 (13 .6-7,334.4)366.8 (1 1 9-1 ,073 .9}262.8 (1 0.1-30,534.0)358.6 (1 0.1-3,538.5}250.9 (1 0.8-30,534.0)

OBSTETRICS & GYNECOLOGY


6/9

I

I

l)

Table 4. Stratification of Pe lvic Mass Patie nts With Benign Disease and Can ce r Based on the Ris k ofOvarian Malignancy Algorithm

Pre menopausalDisease Status lo w Risk High RiskBenign 173 60low malignant potential tumor 3 4EOC stage 1-11 0 JEOC Stage III-IV 0 5EOC unstaged 0 1Non-COC 0 0Other gynecologic cancers 1 2Metastatic cancers 1 1Other 1 0Total 179 76EOC, epithelial ovarian cancer.

55.4-88.1%), a specificity of 74.9% (95% CI 70.3-79.2%} , and a negative predictive value of 97.3% (95%CI 94.7- 98.8%).

Examination of all women with benign neoplasms (n=383) or with an y malignancy includinglow malignant potential tumors {n =89}, the Risk ofOvarian Malignancy Algorithm correctly classified 72of 89 malignant or low malignant potential tumorsachieving a sensitivity of 80.9% (95% CI 71.2-88.5%)at a specificity of 74.9% (95% CI 70.3- 79.2%). Inpostmenopausal women only, the Risk of OvarianMalignancy Algorithm correctly classified 56 of 67malignant or low malignant potential tumo rs achievin g a sensitivity of 83.6% (95% CI 72.5-91.5%) at aspecificity of 76.0% (95% CI 68.4-82.6%) and inpremenopausal women only, the Risk of OvarianMalignancy Algorithm correclly classified 16 of 22malignant or low malignant potential tumors achievin g a sensitivity of 72.7% (95% CI 49.8-89.30/o) a laspecificity of 74.2% (95% CI 68.1-79.7%).

Postmenopausal Co mbinedlo w Risk High Risk lo w Risk High Risk

114 36 287 962 10 5 143 6 3 90 29 0 340 1 0 21 1 13 5 4 70 0 1 12 4 3 4125 92 304 168

DISCUSSIONThe serum biomarker HE4 is a novel markerrecently cleared by the U.S. Food and Drug Ad ministration for ovatian cancer monitoring. HE4 isa member of a family of protease inhibitors thatfunction in protective immunity an d consists of twowhey acidic protein domains an d a 4-disulfide core.The HE4 gene is expressed by epithelial ovariantumors an d ca n be detected through immunohistochemical staining of malignant ovarian tissue. 17- 20 TheHE4 protein expressed by epithelial ovarian malignancies ca n also be detected in the serum of patientswith ovarian cancer.21 Equally important, IIE4 hasbeen shown to be a sensitive marker for the differentiation of benign ovarian tumors from those that ar emalignant and a marker for adenocarcinomas of theendometrium. 13141621 Using area under the receiveroperator characteristic curves {AUC ROC) as a measure of test performance, a pilot study examining nine

Ta ble 5. Distribution of Patie nts Int o Low-Risk an d High-Risk Groups: Be nign Compa red With EpithelialO va rian Cancer a nd Low Malignant Pote ntial Tu mo rsMenopausal Status Disease lo w Risk* High Risk*

Combined Benign 287 (75) 96 (25)Cancer 8 (12) 59 {88)Total 295 (66) 155 (34)Premenopausa I Benign 173 (74) 60 (26)Cancer 3 (19) 13 (81)Total 176 (71) 73 (29)Postmenopausal Benign 114 (76) 36 (24)Cancer 5 (10) 46 (90)Total 119 (59) 82 (41 )PPV, positive predictive value; 1\PV, negative predictive value.Data are n (%) unless otherwise specified.

Total383674502331624q

15051201

Percentage of cases within the low-risk group and within the high-risk group .

Sensitivity Specificity PPV NPV88.1 74.9 38.1 97.3

81.3 74.2 17.8 98.3

90.2 76.0 56.1 95.8

VOL. 118, 0. 2, PART 1, AUGUST 2011 Moore et al Use of the Risk of Ovarian Malignancy Algorithm 285


7/9

Table 6. Percentage of Epithelial Ovarian Cancer Misclassified to the low-Risk Group and CorrectlyClassified to lhe High-Risk GroupMenopausal EOC Stage Total Percent EOC Percent EOCGroup lMP I and II Ill and IV Unstaged EOC Misclassified Co rrectly ClassifiedPremenopausal 3 0 0 0 9 0 100Postmenopausal 2 3 0 0 19 7.7 92.3All patients 5 3 0 0 48 6.2 93.8LMP, low malignant potential tumors; [ 0 ( , epithelial ovarian cancer.

potential serum biomarkers for ep ithelial ovariancancers found HE 4 to be the mosl sensitive markerfor ovarian cancer. More importantly, it was demon-strated that a combination of the serum biomarkersHE4 and CA 125 achieved a higher AU C ROC andtherefore had increased sensitivities than eithermarker alone. These findings have since been independently validated. Huhtincn et al demonsiYated thecombination of serum HE4 and CA 125 can differentiate endometriosis from an ovarian malignancy witha sensitivity of 79% compared with CA 125 and HE4alone \'lith sensitivities of 64% and 71%, respectively.As well, when studying ovarian cancers compared\'lith controls, the dual marker combination achieveda sensitivity of 93% compared with CA 125 or HE4,which each achieved a sensitivity of 79% alonc.22More recently, Nolen ct al evaluated 65 ovariancancer-related biomarkers in the circulation ofwomen diagnosed with an adnexal mass and foundthat as individual markers, HE4 and CA 125 providedthe greatest level of discrimination between benignand malignant cases. In addition, consistent \'lith ourprior trial, these researchers also demonstrated thatthe combination of the biomarkers IIE4 and CA 125provided a higher level of discriminatory power thaneither marker considered a1one.23

The stratification of women into menopausalgroups is critical because biomarker expression, bothH4 and CA 125, can vary depending on age and thepresence of benign or malignant tumors in the twogroups. Therefore, these variables must be taken intoaccount when developing algorithms using biomarkers. The Risk of Ovarian Malignancy Algorithm wasdeveloped from lhe combination of two pilot studiesand uses serum levels of HE4 and CA 125 along withmenopausal status to stratify patients into high-riskand low-risk groups for epithelial ovarian cancer. 1314The Risk of Ovarian Malignancy Algorithm was thensubsequently validated in a multicenter trial assessingwomen who presented with a pelvic mass. The cohortof women in lhis trial was considered to be atincreased risk for having a malignancy because allpatients were initially evaluated and enrolled by

gynecologic oncolot,rists. The initial validation nialhad an incidence of all cancers of 33% and 240fo forepithelial ovarian cancer alone for the women enrolled onto the trial. With this in mind, the applicationof the Risk of Ovarian Mal ignancy Algorithm to thehigher-risk cohort of women separated patients intohigh-risk and low-risk groupl> effectively. When assessing premenopausal and postmenopausal womentogether, the Risk of Ovarian Malignancy Algorithmstratified women with benign compared with epithelial ovarian cancer and low malignant potential tumors into high-risk and low-risk groups with a sensitivity of 89%, a specificity of75%, and a 94% negativepredictive value. When assessing postmenopausalwomen alone, the Risk of Ovarian Malignancy Algorithm achieved a sensitivity of92% and a specificity of75% with a negative predictive value of 93%. Forpremenopausal women alone, the Risk of OvarianMalignancy Algorithm achieved a sensitivity of 77%and a specificity of 75% with a negative predictivevalue of 95%. More importantly, the Risk of OvarianMalignancy Algorithm detected 94% of the womenwith an invasive epithelial ovarian cancer and 85% ofthe women with early-stage epithelial ovarian cancerin the initial validation trial. The findings of a decreased sensitivity for the premenopausal group canbe accounted for by a higher incidence of low malignant po tential tum ors that do not overexpress CA 125or HE4 and the increased incidence of benign tumorsthat overexpress CA 125 in lhe premenopausal givingrise to false-positive tests.

In the currenl trial reported here, all patients wereinitially evaluated by gynecologists, family practitioners, internists, ga.stroenterologists, or general surgeons. For this cohort, the incidence of all cancers was150fo and for epithelial ovarian cancer alone, it was10%. These rates are significantly lower than in thepreviously reported validation trial and more in linewith the population of women being assessed bygeneralists and primary care physicians. Despite thelower incidence of cancer in this study, the Risk ofOvarian Malignancy Algorithm stratified women intohigh-risk an d low-risk groups with similar sensitivities

286 Moore et al Use of the Risk of Ovarian Malignancy Algorithm OBSTETRICS & GYNECOLOGY


8/9

and specificities as seen in the prior validation trial.The high sensitivity achieved through the use of theRisk of Ovarian Malignancy Algorithm allows for94% of women with an epithelial ovarian cancer to beidentified and triaged to a gynecologic oncologist,whereas 75% of the patients with a pelvic mass will betreated by their gynecologist in their community.Equally important is the high negative predictivevalue (97%) achieved by the Risk of Ovarian Malignancy Algorithm for women with epithelial ovariancancer and low malignant potential tumors. This highnegative predictive value provides a strong reassurance that a pelvic mass is benign. Accurate detectionof women at low risk for malignancy would reduceunnecessary referrals to a gynecologic oncologist allowing the patient to stay in their community withtheir primary gynecologist and support network witha low risk of ultimately being diagnosed with amalignancy.

Traditionally, the biomarker CA 125 has beendemonstrated to be elevated in only half of patientswith early-stage ovarian cancer and elevated in up to80-90% of all patients with ovarian cancer.24.25 Therefore, the use of biomarkers for the detection ofearly-stage disease has been limited. Examining patients with stage I and ll disease, the Risk of OvarianMalignancy Algorithm stratified the majority (75%) ofpatients with early-stage ovarian cancer correctly tothe high-risk group, therefore identifying preoperatively patients who will benefit from full surgicalstaging, which will dictate subsequent managementand lhe need for chemotherapy.Recently, lhe Risk of Ovarian Malignancy Algorithm has been evaluated and validated in populations outside of lhe United States. Montagnana et alreported on their fmdings on the performance of HE4and CA 125 for detecting ovarian cancer. Using anAUC ROC analysis, these researchers found thatHE4 achieved higher AUC ROC compared with CA125 alone (AUC, 0. 77 compared with 0.64) for premenopausal women and (AUC ROC, 0.94 and 0.84)for postmenopausal women, similar to the findingdescribed in our pilot trial.26 Likewise, when combining HE4 and CA 125 and using the Risk of OvarianMalignancy Algorithm in premenopausal and postmenopausal women separately, the authors reportthat the Risk of Ovarian Malignancy Algorithm ha dan AUC ROC of 0.77 compared with an AUC ROCof 0.64 for CA 125 alone in the premenopausal groupand a AUC RO C of 0.92 for the Risk of OvarianMalignancy Algorithm and 0.84 for CA 125. Interestingly, the AUC RO C of HE4 compared with that ofthe Risk of Ovarian Malignancy Algorithm was near

equal in bolh the premenopausal and postmenopausal groups.26 The size of this study was smallcompared with previously reported trials and may nothave had the power to detect a dillerence between theuse of CA l 25 and HE4 over that of HE4 alone. Kimet al studied the use of the Risk of Ovarian Malignancy Algorithm in a Korean population and foundthat the Risk of Ovarian Malignancy Algorithm classified women into high-risk and low-risk groups witha sensitivity of 88% and a specificity of 94%, findingsthat are consistent with our two multicenter trials.27

The currenl lrial reported in this issue validatesthe previously reported trial for the use of the Risk ofOvarian Malignancy Algorithm to stratify patientsinto risk groups. Despite a lower incidence of epithelial ovarian cancer in this study population, the Riskof Ovarian Malignancy Algorithm performed equallyas well, attaining high sensitivities and specificities forthe prediction of a women having ovarian cancer.There is an urgent need for tools to help physiciansidentify which women with an ovarian cyst or pelvicmass are al high risk for a malignancy to accuratelytriage palienls to surgeons and institutions specializingin the care and management of ovarian cancer whileretaining mosl of the women with benign disease formanagement by their gynecologist The accumulatingdata support the use of HE4 and CA 125 along withthe Risk of Ovarian Malignancy Algorithm as anaccurate tool to assist in the triage of women "-\lith apelvic mass.REFERENCESJ. American Cancer Society. Cancer facts & figures 2010. Atlanta(GA): Americnn Cancer Society; 2010; p. 1-62.

2. Bristow RE, PaliB BE, Chi DS, Cliby WA . The NationalCancer Da.tabase report on advanced-stage epithelial ovariancancer: impact of hospital surgical case volume on overallsurvival and surgical treatment paradigm. Gynecol Oneal2010;118:262-7.3. Earle CC, Schrag D, Neville BA, Yabroff KR, Topor M, FaheyA, et al. Effect of surgeon specialty on processes of care andoutcomes for ovarian cancer patients. J Nad Cancer lnst2006;98: 172 80.4. Giede KC, Kieser K, DodgeJ, Rosen B. Vvho should operateon patients with ovarian cancer? An evidence-based review.Gynecol Oneal 2005;99:447-61.5. Mercado C, Zingmond 0, Karlan BY. Sekaris , Gross J,Maggnrd-Gibbons M, et al. Quality of care in advancedovarian cancer: the importance of provider specialty. GynecolOneal 2010;117:18-22.6. Vemooij F, Heintz P, Witteveen , van der Graaf Y. Theoutcomes of ovarian cancer treatment are better when provided by gynecologic oncologtsts and in specialized hospitals:a systematic review. Gynecol Oncol2007;105:801-12.7. Guidelines for referral to a gynecologic oncologist: rationaleand benefits. 1'he Society of Gynecologic Oncologists. Gynecol Oncoi2000;78:Sl - l3.

VOL. 118, NO. 2, PART 1, AUGUST 2011 Moore et al Usc of the Risk orOvarian Malignancy Algorithm 287


9/9

8. Th e role of the generalist obstetrician-gynecologist in the earlydetection of ovarian cancer. ACOG Committee Opinion No.280. American College of Obstetricians an d Gynecologists.Obstet Gynecol 2002;100:1413-6.!J. Engelcn MJ, Kos HE, Willemse PH, AaldersJG, de Vries EG ,Schaapveld M, ct al. Surgery by consultant gynecologic oncologists improves survival in patients with ovarian carcinoma.

Cancer 2006 Feb 1; 106:589-98.10. Goff BA, Mallhews BJ, Larson EH, Andrilla CH, Wynn M.Lishner OM, et al. Predictors of comprehensive surgical treat

ment in patients witn ovarian cancer. Cancer 2007;109:2031-42.l l. Paulsen T, Kjaerheim K, KaernJ, Tredi S, Trope C. Improvedsborl-lcrm survival for advanced ovarian, tubal, and peritonealcancer p a t i e n t . ~ operated at teaching hospitals. In t J GynecolCancer 2006;16(suppl 1):11-7.12. Veroooij .F, Heintz AP, CoeberghJW, Massuger LF, WitteveenPO, van de r Graaf Y. Specialized and high-volume care leadsto better outcomes of ovarian cancer treaunent in the !'\etherlands. Gynecol Oncol2009;112:455-61.13. Moore RG, Brown AK, Miller MC, SkatesS, Allard \\J, Verch

T, el al. TI1e use of multiple novel rumor biomarkers for thedetection of ovanan carcinoma Il l patients with a pelvic mass.Gynecol Oncol 2008;108:402- 8.14. Moore RG, McMeekin DS, Brown AK, Disilvestro P, Miller

MC, Allard WJ, ct al. A novel multiple marker bioassayutilizing HFA and CA12.5 for the prediction of ovarian cancerin patients with a pelvic mass. Cynecol Oncol2009;112:40-6.

15. Moore RC , Jabre-Raughley M, Brown AK, Robison KM,Miller MC, Allard \ ~ , et al. Comparison of a novel multiplemarker assay vs the Risk of Malignancy Index for the prediction of epithelial ovar ian cancer in patients with a pelvic mass.AmJ Obstet Cynccol 2010;203:228.e1-6.

16. Moore RG, Brown AK, Miller MC, Badgwell D, Lu Z, Allardet al. Util ity of a novel serum tumor biomarker HE4 inpatients with endometrioid adenocarcinoma of the uterus.Cynccol Oncol2008;1 10: 196-201.

17. Drnpkin R, von Horstcn I lH , Lin Y, Mok SC, Crum CP,Welch WR, et al. Human epididymis protein 4 (HE4) is asecreted glycoprotein that is overexpressed by serous andendometrioid ovarian carcinomas. Cancer Res 2005;65:2162- 9.

18. Calg-.mo ~ I T . Hampton CM, Frierson HF Jr . ComprehensiveanalysiS of l l ~ expression in normal and malignant humantissues. Mod Pathol 2006;19:847- 53.

19. Hough CD , Shcnnan-Baust CA, Pizer ES, Montz FJ, Im DD,Rosenshein NB, et at. Large-scale serial analysis of geneexpression reveals genes dilTerentially expressed in ovariancancer. Cancer Res 2000;60:6281-7.20. WelshJn, Zarrinkar PP, Sapinoso LM, Kern SC, Behling CA,Mon.k BJ, el al. Analysis of gene expression profiles in normal

and neoplastic ovarian tissue samples identifies candidatemolecular markers of epithelial ovarian cancer. Proc Nat! AcadSci US A 2001;98: 1176-81.21. Hellstrom I, Raycraft], Hayden-Ledbetter M, Ledbetter JA ,

Schummer M, Mcintosh M, et al. Th e HE4 {WFDC2) protemis a boomarker for ovar1an carcinoma. Cancer Res 2003;63.3695-700.

22. lluhtinen K, Suvitie P, HiissaJ,JunnilaJ, HuvilaJ, Kujari H,ct al. Serum HE4 concentration differentiates malignant ovarian rumours from ovarian endometriotic cysts. Br J Cancer2009;100:1315- 9.

23. Nolen H, Velikokhatnaya L, Marrangoni A, De GK. LomakinA, Bast RC Jr , et al. Serum biomarker panels for the discrimmation of benign from malignant cases in patients with anadnexal mass. Gynecol Oncol 2010;117:440-5.

24. Bast RCJr, KJug TL, Stjohn E,Jeruson E, NiloffJM, LazarusII, et al. A radioimmunoassay using a mono clonal antibod y tomonitor the course of epithelial ovarian cancer. N EngiJ Med1983;309:883-7.

25. Woolas RP, Xu !J,Jacobs1], Yu YH, Daly L, llerchuckA, elat. Elevation of multiple serum markers in patients with stage1 ovarian cancer J Nat! Cancer lust 1993;85:1748-,Sl.26. Montagnana M, DaneseE, Ruzzenente 0, BrescianiV, Nuzzo

T, Gelati M, ct al. The ROMA (The Risk of Ovarian Malignancy Algorilhm) for estimating the risk of epithelial ovariancaJ.lcer in women presenting with pelvic mass: is it reallyusefu l? Clin Chern Lab Med 2011 ;49:521-5.27. Kim YM. Whang DJT, Park.J, Kim SH, Lee SW, Park HA, etal. Evaluation of the accuracy of serum human epididymisprotein 4 in combination with CA125 for detecting ovariancancer: a prospective case-control study in a Korean population. Chn Chem Lab Med 2011;49:527-34.

288 Moore et al Use of he Risk of Ovarian Malignancy Algorithm OBSTETRICS & GYNECOLOGY

green journal roma

Documents