A multidisciplinary approach to the early detection of ovarian carcinoma: Rationale, protocol design, and early results

A multidisciplinary approach to the early detection of ovarian carcinoma: Rationale, protocol design, and early results Beth Y. Karlan, MD; Leslie J. Raffel, MD,b Gordan Crvenkovic, MD; Cindy Smrt, RT, RDMS; M. Dwight Chen, MD; Eloise Lopez, RN, BSN; Catherine A. Walla, RN, MA, MN; Carrie Garber, MS,b Patricia Cane, PhD; Dennis A. Sarti, MD; Jerome I. Rotter, MD,b and Lawrence D. Platt, MD"

Los Angeles, California OBJECTIVE: This study was undertaken to determine the feasibility of currently available tests to detect ovarian cancer at a curable stage. By studying women at risk for the disease on the basis of their family histories of cancer, we hope to gain insight into the genetics and biologic characteristics of ovarian cancer. STUDY DESIGN: Asymptomatic women with a family history of cancer are interviewed by geneticists to obtain accurate pedigree and epidemiologic data. Screening tests including transvaginal ultrasonography, color Doppler imaging, CA 125 measurement, and other investigative biomarkers are performed between cycle days 3 and 8 and repeated biannually. RESULTS: A total of 597 women without symptoms were screened between July 1991 and June 1992. Cystic or complex adnexal masses were detected in 6.2% of patients. A pulsatile index value < 1.0 was measured in 80% of premenopausal patients and 24% of postmenopausal patients, whereas a resistance index value < 0.4 occurred in 12% of premenopausal and 3% of postmenopausal patients. A CA 125 level > 35 U/ml was found in 11.4% of the study participants. To date, one stage IA borderline ovarian tumor and one endometrial carcinoma have been found. CONCLUSIONS: Ovarian cancer screening needs to be investigated in a controlled fashion to determine means that will ultimately improve the survival from the disease. (AM J OSSTET GVNECOL 1993; 169:494-501.)

Key words: Ovarian cancer screening, transvaginal ultrasonography, color Doppler imaging, CA 125, familial cancers

Ovarian cancer continues to be the leading cause of death from gynecologic malignancy and will be responsible for the deaths > 100,000 U.S. women over the next 5 years. For perspective, more U.S. women are killed annually by ovarian cancer than by homicide and suicide combined. I Continued difficulties in the early detection of ovarian carcinoma have resulted in almost two thirds of the patients having widespread metastatic From the Department of Obstetrics and Gynecologya and the Division of Medical Genetics, Departments of Pediatrics and Medicine, b Cedars-Sinai Medical Center, University of California, Los Angeles, School of Medicine. Supported by the American Cancer Society, Career Development Award (B.Y.K.); Dianon Systems Inc., Stanford, Connecticut; Advanced Technology Laboratories, Bothell, Washington; Acuson Computed Sonography, Mountain-View, California; The Helping Hand of Los Angeles; and the Stuart Foundations and the Cedars-Sinai Board of Governors' Chair in Medical Genetics (j.l.R.). Received for publication November 30, 1992; revised March 26, 1993; accepted April 13, 1993. Reprint requests: Beth Y. Karlan, MD, Cedars-Sinai Medical Center/UCLA, Department of Obstetrics and Gynecology, 8700 Beverly Blvd., Los Angeles, CA 90048. Copyright © 1993 by Mosby-Year Book, Inc. 0002-9378/93 $1.00 + .20 6/1/47913

494

disease at initial diagnosis. The outlook for these women with advanced disease remains grave with no more than 15% of them living > 5 years! When ovarian cancer is diagnosed in its early stages, however, the prognosis for patients is less grim, with 5-year survival rates approaching 95% reported for women with stage I disease. 3 • 4 Although new therapies continue to be sought, an effective means of early detection could have an even greater impact on improving survival from this disease. The various difficulties in routinely screening women without symptoms for ovarian cancer have been the subject of many recent journal articles and conferences. 5 -7 The relatively low prevalence of epithelial ovarian cancer, coupled with our inability to define a precursor lesion and to determine the lag time for development of widespread disease, have led to concerns about the feasibility of screening for this often lethal disorder. The currently available screening modalities have not demonstrated sufficient sensitivity or specificity to provide a reliable means to detect early ovarian cancer. Even if effective, the cost of these

Volume 169, Number 3 Am ] Obstet Gynecol

screening tests, if applied to the general population, would account for a significant portion of our health care budget. 7 This disheartening but realistic analysis of ovarian cancer screening in 1993 is coupled with a heightened public awareness of the disease, since the untimely death of comedienne Gilda Radner in 1989 and the increased media coverage depicting the "plight" of patients with ovarian cancer. This education of U.S. women about ovarian cancer has had many benefits and has probably resulted in the earlier presentation and treatment of some patients. A paranoia has arisen, however, because of the frequently silent growth of these tumors and our present inability to effectively screen for this disease in its most curable stages. Patients are demanding reassurance from screening tests such as CA 125 measurement that have an unproven efficacy in this arena. This fervor has been fed by "ovarian cancer screening centers" offering fee-forservice tests and a false sense of security to a vulnerable female patient population. We believe that ovarian cancer deserves careful study to determine if an effective means of early diagnosis can be found. Ultimately, this intervention must also be shown to improve survival from the disease. With these tenets in mind, we have assembled a multidisciplinary team to screen a population of women at increased risk for ovarian cancer on the basis of their families' cancer histories. Our rationale and protocol design are described here. Genetic screening

A family history of ovarian cancer is the strongest known risk factor for development of the disease. This information is supported by many epidemiologic studies,8-13 which point out the diversity of heritable ovarian tumors. There are at least five rare genetic syndromes associated with ovarian tumors such as gonadal dysgenesis (with an XY germ cell line), Peutz-Jeghers syndrome, allier's disease, the basal cell nevus syndrome, and familial ovarian fibromatosis. 14 Our focus has been on the more common epithelial ovarian carcinomas where at least three pedigree patterns have been described: familial "site-specific" epithelial ovarian cancer, familial ovarian cancer associated with breast carcinoma (the "breast-ovary syndrome"), and familial ovarian cancer as part of the Cancer Family or Lynch II syndrome. This latter syndrome is characterized by inherited nonpolyposis colon cancer or endometrial cancer, or both, as well as by ovarian cancer in affected family members. 15. 16 Extended pedigrees have been reported for each of these disorders and suggest that the genetic susceptibility to ovarian cancer may be inherited in an autosomal dominant fashion with incomplete penetrance. The risk therefore can be trans-

Karlan et al.

495

mtited through either the maternal or the paternal lineage. These different syndromes indicate the presence of genetic heterogeneity in ovarian cancer inheritance, as a result of either locus heterogeneity (the involvement of entirely different genes) or allelic heterogeneity (different mutations at the same locus giving rise to different phenotypes), or both. It is estimated that < 5% to 10% of epithelial ovarian cancers have a hereditary component. However, the empiric risk for ovarian cancer is also increased for women with firstdegree relatives with breast cancer, colon cancer, and possibly endometrial cancer, even in the absence of a pedigree suggestive of autosomal dominant inheritance. 9. 16 On the basis of these data, eligibility for our study requires that patients have a first-degree relative (parent, sibling, or offspring) with ovarian, breast, endometrial, or colon cancer or a second-degree relative (grandmother, granddaughter, or aunt) with ovarian cancer. Ultrasonography and Doppler studies

There has been great int~rest and enthusiasm for the role of ultrasonography as an ovarian cancer screening modality.17-20 The ovaries can be visualized in > 95% of premenopausal patients and up to 85% of postmenopausal patients. 18 When the ultrasonographic characteristics of the ovary are compared with the gross ovarian morphologic features at the time of surgery, the correlation is between 78% and 99%. The accuracy of ultrasonography in predicting the pathologic diagnosis, however, is only between 30% and 68%.17 The seminal studies by Van Nagell et al. of transvaginal ultrasonography for screening women without symptoms for ovarian cancer have yielded sobering results. No ovarian carcinomas were detected in the first 1000 women screened, although other pathologic conditions were detected (e.g., metastatic colon cancer, serous cystadenoma, endometrioma, hydrosalpinx, etc.).18 A subsequent study by this group, focusing on 1300 postmenopausal women without symptoms, discovered two early stage ovarian cancers among 33 women with detectable ovarian ultrasonographic abnormalities (27 of whom underwent laparotomy). 19 Recognizing the ability of transvaginal ultrasonography to accurately portray ovarian morphologic features, we sought means to improve its diagnostic accuracy for ovarian carcinoma. Color Doppler flow imaging provides a noninvasive method to evaluate blood flow patterns, and this technique can easily be coupled with real-time transvaginal ultrasonography. Neoangiogenesis is an obligate early event in tumor growth. These new tumor vessels have ultrasonographic characteristics distinguishable from normal vessels by color Doppler studies. 21 , 22 Initial experience with color Doppler flow studies of ovarian masses has demonstrated a high

496 Karlan et al.

degree of specificity for distinguishing benign from malignant tumors. 23 - 25 Therefore we sought to apply this technique to the screening of women without symptoms in an attempt to increase our diagnostic accuracy for early ovarian cancers. Biomarker analysis

Although a "simple blood test" for the diagnosis of ovarian cancer has long been sought, the complexity of the disease and our poor understanding of its basic biologic characteristics have kept this potentially useful screening modality elusive. Over the last 10 years, multiple tumor markers have been detected and studied in patients with ovarian cancer. CA 125 is the best characterized marker and has demonstrated great clinical utility in the monitoring of patients with known ovarian carcinoma. 26. 27 Elevations of CA 125 levels have been reported to precede the diagnosis of stage III ovarian cancer by 10 to 12 months!8 As a screening tool, however, CA 125 has not demonstrated sufficient specificity or sensitivity for the general population. CA 125 elevations are noted in only 23% to 50% of stage I ovarian cancer;9 yet frequently CA 125 elevations accompany a long list of benign disorders. Einhorn et al. 30 recently reported on 5550 apparently healthy women screened with CA 125 measurement and reported a specificity of 98.5% (35 Ulml cutofI) for women > 50 years old. They suggested that a much larger study would be appropriate in this group of women to determine the impact of CA 125 screening on survival of patients with ovarian cancer. Clearly the role of CA 125 as a screening test has not yet been defined. In an attempt to improve on the specificity and sensitivity of CA 125, four additional tumor markers were added to our screening regimen. Lipid-associated sialic acid has been found to be a marker for malignant disease on the basis of elevated levels of sialoglycoconjugates associated with the tumor's growth. 3l Although relatively nonspecific for ovarian cancer, it was thought to be a useful component in our multiple marker profile, which includes more type-specific biomarkers. The marker DF!70K was extracted from epithelial ovarian tumor tissue and can be found in patients with epithelial ovarian tumors of all histologic types, including mucinous. This marker, which is immunologically distinct from CA 125, is also a good adjunct to CA 125. 32 Urinary gonadoprotein has proved a sensitive marker for gynecologic malignancies with specificities > 90% in patients with ovarian cancer.33 Amplification of the HER-21neu oncogene has been associated with poor survival in patients with ovarian cancer whose tumors overexpress this oncogene. 34 HER-2neu encodes a transmembrane glycoprotein that is shed into the sera of some patients with ovarian cancer. The potential screening role of this new marker is also being investigated in our program.

September 1993 Am J Obstet Gynecol

Epidemiologic factors and serum, urine, and tissue banking

In addition to the screening for ovarian cancer, we recognized the opportunity to study this population with respect to their exposure to environmental factors and their reproductive and medical histories. The collection of family histories for assessment of eligibility enables us to evaluate the impact of genetic factors on risk. By collecting specimens from patients with significant family histories of cancer, we can store materials for future studies on women who may carry a gene for ovarian cancer susceptibility. Therefore, for patients and family members in whom autosomal dominant transmission is suspected, lymphoblastoid lines are developed and stored and tumor tissues are obtained as availability permits. In related studies in our laboratories, these tissues are being investigated. During the longitudinal follow-up of study patients new diagnostic modalities may become available, so serum and urine from all participants are banked biannually. Material and methods

Information regarding our program is disseminated by educational lectures, physician letters, and media publicity. The dedication and naming of our early detection study for Gilda Radner probably contributed to the initial brisk response and continued public interest in the program. Patients are accrued from the community through physician referral and self-referral. Telephone screening confirms that the women are without symptoms, > 35 years old, available for longitudinal follow-up over the next 5 years, and have a family history of at least one first-degree relative with breast cancer, colon cancer, ovarian cancer, or other gynecologic cancer or a second-degree relative with ovarian cancer. Informed consent is obtained from each patient before entry into this institutional review boardapproved ovarian cancer screening study. Patients are asked to complete a family history questionnaire and obtain pathology reports on affected members' tumors before their initial visit. They meet individually with a genetic counselor to review their risk for ovarian cancer. Epidemiologic questionnaires are also completed and reviewed (Table I). Vaginal ultrasonography and Doppler studies are performed by four experienced research sonographers with 5 to 20 years' experience in the departments of obstetrics and gynecology and radiology. An ATL Ultramark 9-HDI unit or an Acuson 128XP/10 ultrasonography unit, each equipped with a 5 MHz vaginal transducer, is used. All patients have an empty bladder and are scanned in a lithotomy position. Each ovary is measured in three planes to determine the transverse (D]), anteroposterior (D 2), and longitudinal (D3) diameters. Ovarian volumes are calculated with the prolate ellipsoid formula: Volume = 'rI'/6 x D] x D2 X D3.

Karlan et al.

Volume 169, Number 3 Am J Obstet Gynecol

497

Table I. Principal categories included in family history and epidemiologic questionnaire Demographic characteristic

Reproductive history

Medical and surgical history

Detailed family pedigree

Height and weight Ethnicity Religion Occupation Education Habits Alcohol use Tobacco use Recreational drug use Diet (% fat)

Gravidity Parity Birth control Hormone replacement therapy Prior gynecologic abnormality or operation Infertility

Active disease Mumps Breast disease Other cancers Medications

Multiple generations Maternal Paternal Cancer histories Any known genetic diseases

Ovarian architecture is classified on the basis of the echogenicity of cystic, solid, and complex components, A systemic evaluation of the uterus is also made and includes measurements of endometrial thickness and the presence of pelvic fluid, Quality control studies have been demonstrated by means of intraobserver and interobserver reproducibility in the ultrasonographic measurements. Blood flow is visualized and flow velocity waveforms are obtained from the right and left ovarian arteries and the internal iliac artery, as well as the endometrium and uterine artery. Additionally, blood flow is measured at multiple sites on each ovary, including central, peripheral, pericystic, and adjacent vessels. A minimum of three identical velocity waveforms are used to calculate the resistance and pulsatility index values defined by the following equations: Resistance index = (Peak systolic velocity - Peak diastolic velocity)!(Peak systolic velocity); Pulsatility index = (Peak systolic velocity Peak diastolic velocity)/(Mean velocity). Premenopausal patients are scanned between days 3 and 8 of the cycle, and any abnormalities that are found are confirmed by another sonographer during the subsequent cycle. For these purposes persistent adnexal masses and/or resistance index values < 0.4 are considered abnormal. Follow-up transvaginal ultrasonography and Doppler examination are routinely scheduled at 6-month intervals. At each scheduled appointment, blood and urine are obtained from patients for biomarker analysis. Biomarker studies are repeated in subsequent cycles when significant abnormalities are found. Serum and urine from each patient are banked for future analysis of new tumor markers, and lymphoblastoid lines are developed on appropriate patients and their available family members for future deoxyribonucleic acid analysis. Suppression of ovulation with low-dose oral contraceptives is suggested to premenopausal patients and their physicians if their family history suggests a significant increase in ovarian cancer risk. All services are provided to patients at no cost as a part of this research study protocol.

Results

Between July 1, 1991 and June 30, 1992, >3000 telephone calls were received from patients interested in our ovarian cancer detection program. The onslaught of inquiries from the public expressing interest in our screening program speaks to the level of anxiety and concern regarding the early detection of ovarian cancer. Appointments for protocol entry and study are being filled> 6 months in advance. Telephone inquiries are made by both male and female members of potentially affected families; an urgency in their requests for appointments and a sense of relief after study preparation has been the rule. The psychologic impact of ovarian cancer screening is also being studied with the use of questionnaires on follow-up visits to assess the self-reported changes in anxiety levels and information retention since entry in the program. A total of 597 women met eligibility criteria and consented to be participants in the screening program. Four hundred forty-nine premenopausal women (75.2%) and 148 postmenopausal women (24.8%) were seen in the first year of our study. Eight hundred sixty-three vaginal ultrasonography examinations and Doppler studies were performed, and 887 biomarker panels were measured. Table II outlines the distribution of patients by their family histories. More than half of the patients entered our program because a single first- or second-degree relative had ovarian cancer. However, 76 (12.8%) had a sufficient number of family members with cancer and pedigree structures that were thought to be consistent with a hereditary cancer syndrome. Ovaries were visualized in 442 of the premenopausal women (98.4%) and 135 of the postmenopausal participants (91.2%). The high rate of ovarian identification, especially in the postmenopausal group, may have been due to (1) the use of color flow imaging, which allows easy identification of the external iliac vessels and thus the neighboring ovary, or (2) operator experience, or both. At the time of the initial visit, the mean ovarian volume of premenopausal patients was 8.50 ± 10.38 cm 3 (right) and 9.25 ± 14.03 cm 3 (left); for postmeno-

498

Karlan et al.

September 1993 Am J Obstet Gynecol

Table II. Distribution of patients by their family histories Family history

No.

%

First- or second-degree relative with ovarian cancer First-degree relative with gastrointestinal, breast, or gynecologic cancer Patient with breast cancer Site-specific ovarian cancer syndrome Breast or ovary cancer syndrome Lynch II cancer syndrome Dominant cancer pedigree* Ineligible patients

336 161 18 38 13 7 18 6

56.3 27.0 3.0 6.3 2.2 1.2 3.0 1.0

TOTAL

597

100.0

*Dominant cancer pedigree where ovarian cancer is not a defined cancer site (e.g., breast or colon.)

pausal patients the values were 3.85 ± 4.24 cm 3 (right) and 4.85 ± 6.9 cm 3 (left). Cystic adnexal masses (~5 cm) were observed in 4 premenopausal patients (0.7%) and none of the postmenopausal women; complex masses were identified in 7 premenopausal women (1.2%) and 3 postmenopausal women (0.5%). The color Doppler flow data revealed pulsatility index values < 1.0 in 359 premenopausal women (80%) and 36 postmenopausal patients (24%); pulsatility index values> 1.0 are the reported normal range in the literature. 23 Because of the very large number of normal women with a pulsatility index value < 1.0, this screening test has not proved to be very useful in our hands. Resistance index values were :5 0.4, the proposed normal value reported by Kurjak et aJ.2 4 in 54 of our premenopausal patients (12%) and 4 of our postmenopausal patients (3%). This indirect measurement of ovarian blood flow appeared to be a better test. Patients with ovarian resistance index values :5 0.4 were restudied 1 month later. Sixty-eight of the patients (11.4%) had CA 125 values > 35 U/m!' Most of these patients (n = 60) were premenopausal and had ultrasonographic findings consistent with benign conditions associated with CA 125 elevations (endometriosis, adenomyosis, leiomyomas) or have been restudied and have falling or stable values. Eight postmenopausal patients (1.4%) had CA 125 values > 35 U/ml and half of these (n = 4) were > 65 Vim!' All of these women had normal ovarian architecture and Doppler flow index values and are being followed up by their primary gynecologists. The other biomarker levels were considered abnormal and determinations were repeated only if they were greater than two times the cutoff value. These included 1 patient (0.2%) with a lipid-associated sialic acid value > 40 mgldl, 6 patients (1.3%) with a DM/70K value > 70 U/ml, 20 patients (3.4%) with a urinary gonadoprotein value > 10 fmol/mg, and 33 patients (5.5%) with a HER-21neu value> 40 Vim!' Follow-up studies were performed on 115 of the patients (19.3%) because of abnormal test results. The

indications for repeat testing are outlined in Table III. Despite the large number of women with one or multiple screening test abnormalities, only 19 participants have undergone oophorectomy at this time (Table IV). All operations were performed by the patients' private· gynecologists. One tumor of low malignant potential was discovered on the basis of the transvaginal ultrasonographic morphologic characteristics of complexity and multiple septations. The pulsatility index was 0.6 on the left ovary and 0.75 on the right one. The resistance index was 0.4 on the left ovary and 0.51 on the right one. The CA 125 level in this patient was 14 Vim!. Four oophorectomies were performed after genetic counseling and pedigree analysis identifYing a risk of ovarian cancer that may approach 50%. Five patients elected to undergo oophorectomy on the basis of their own concerns. Persistently abnormal Doppler findings were the indication for laparoscopy or laparotomy in 7 patients, whereas abnormal ovarian or endometrial architecture was the impetus for surgery in 3 patients. Endometrial carcinoma was discovered in a woman without symptoms as a result of our ultrasonographic screening and subsequent workup. One hundred forty-seven of the patients (24.6%) have returned for a 6-month follow-up during the first year of this program. Longitudinal study of this population will be necessary to determine the true false-negative and false-positive rates of these modalities in this patient population. Study retention and compliance strategies include reminder telephone calls and appointment postcard mailings, as well as the initial educational bonds made on study entry. Since we instituted the follow-up postcard mailings, retention participants in our study has approached 90%. The rate of detection of these studies will have to take into account the drop-out of study participants as a result of elective or prophylactic oophorectomy, as well as noncompliance. Comment

The goal of this initial report from our ovarian cancer detection program was to outline our rationale and

Karlan et al. 499

Volume 169, Number 3 Am J Obstet Gynecol

Table III. Patients recalled as a result of abnormal findings

No. of patients recalled Recall as a result of Abnormal ovarian architecture Abnormal resistance index value Elevated CA 125 level Abnormal ovarian architecture and abnormal resistance index value Abnormal ovarian architecture and elevated CA 125 level Abnormal resistance index value and CA 125 level Abnormal resistance index value, abnormal ovarian architecture, and elevated CA 125 level Other

No.

%

115

100.0*

24 15 27 13 7 8 3

20.9 13.0 23.5 11.3 6.1 7.0 2.6

18

15.7

*Total of 19.3% of all patients seen.

Table IV. Patients undergoing oophorectomy during first year of study Impetus for surgery*

Mean age (yr)

Pathologic finding Atrophic ovaries, follicular and corpus luteum cysts, serous cystadenofibroma Thecoma, serous cysts, atrophic ovaries Endometrioma, mucinous cystadenoma, follicular and corpus luteum cysts, serous cystadenofibroma Stage IA ovarian tumor of low malignant potentia\' stage IB and grade 2 endometrial adenocarcinoma, inflammatory inclusion cysts

Dominant family historyt

4

44

Patient choice Color flow Doppler findingst

5

7

49 43

Real-time ultrasonographic findings§

3

54

*A1l CA 125 values were normal. tTwo site-specific ovarian cancer syndromes and two breast-ovary cancer syndromes. tPulsatility index value < 1 and resistance index value < 0.4. §Two complex adnexal masses and increased endometrial thickness.

protocol. In addition, preliminary results from our first year's experience are presented. Subsequent detailed reports of our ultrasonographic, biomarker, and genetic findings are forthcoming. Our initial enthusiasm has been somewhat tempered by the enormity of the task at hand at the relatively low prevalence of ovarian cancer, even in this at-risk population. At this time we are focusing on the longitudinal follow-up of patients currently enrolled in the screening program; concerted efforts are being made toward acquiring the complete medical records of the affected family members in this patient cohort. New patients are being added only after their family pedigree and medical records have been reviewed. The determination of genetic risk for ovarian cancer is multifactorial. We believe that our study participants have benefitted from consultation with the genetics counselors to review their family histories and risk factors. Our psychologic studies in anxiety reduction, however, should help to better define the impact of this counseling. The Doppler flow studies revealed normative data that contrast sharply with the values previously reported in the literature!3-25 In our hands there were

fewer false-positive results for the resistance index (12% premenopausal, 3% postmenopausal) than for the pulsatility index (80% premenopausal, 24% postmenopausal) when cutoff values of 0.4 and 1.0, respectively, were used. Therefore the resistance index served as our primary measure of ovarian blood flow. Possible explanations for why our Doppler data differed from those in prior reports include the population studied consisting of patients without symptoms and the greater sensitivity of the ultrasonography units we used. Blood flow velocities as low as 3 cm/sec could be detected, whereas many of the prior reports used equipment capable of detecting a flow of only 8 cm/sec. Visualization of these smaller vessels may be obscuring the more significant findings, thus leading to false-positive results. We also identified flow patterns at multiple ovarian sites. Again, this closer scrutiny of the ovarian vasculature may have resulted in "false-positive" determinations. Clearly, if Doppler technology holds promise as a screening modality, the measurements and equipment need to be standardized to define normative values. Until the parameters to detect ovarian neoangiogenesis and abnormal blood flow are better defined, Doppler studies for ovarian cancer screening should be performed under a

500 Karlan et al.

research protocol and by experienced sonographers. Real-time vaginal ultrasonography allowed clear visualization of premenopausal and postmenopausal ovaries. The ovarian volumes measured in this population are consistent with the values previously reported in the literature. 18 A detailed report of the ultrasonographic findings on this cohort of nearly 600 women without symptoms is forthcoming. The tumor marker analysis again revealed the inaccuracy of CA 125 alone as a screening modality, especially in premenopausal patients. It is too early to predict the role of the other biomarkers, but at this time they appear to have a secondary role to other screening modalities. It is hoped that an algorithm of biomarker and ultrasonographic data will be found that will allow the early detection of ovarian cancer. Subsequent analysis and follow-up will still be required to determine the impact of this testing on the long-term survival from this disease. It is premature to perform a cost-benefit analysis of this multidisciplinary ovarian cancer screening program at this time. The factors on which we will focus in the future include the potential biologic, economic, and psychologic costs or benefits experienced by patients. Our patients will continue to be a driving force behind our investigations to find better treatment and diagnostic modalities for ovarian cancer. Until a method for the early detection of ovarian cancer is better defined, however, screening for this disease should be done in the arena of clinical research. It seems unlikely at this time that a single test, similar to the Papanicolaou smear for cervical cancer or the mammogram for breast cancer, will be available in the near future. We are also investigating other modalities, such as positron emission tomography with fluorine 18fluorodeoxyglucose for the diagnosis of early ovarian cancer. Additionally, the genetic analysis of patients who may carry a gene placing them at increased risk of the development of ovarian cancer may point to genetic markers for the disease and may allow earlier intervention or, ultimately, prevention.

September 1993 Am J Obstet Gynecol

7. 8.

9. 10.

11. 12.

13. 14. 15. 16.

17.

18. 19. 20. 21. 22.

23.

REFERENCES 1. Boring CC, Squires TS, Tong T. Cancer Statistics 1991. CA Cancer] Clin 1991;41:19-36. 2. Petterson F, ed. Volume 20: annual report on the results of treatment in gynecological cancer. Stockholm: International Federation of Gynecology and Obstetrics, 1990. 3. Gallion HH, van Nagell ]R, Donaldson EG, Higgins RV, Powell DE, Kryscio RJ. Adjuvant oral alkylating chemotherapy in patients with Stage I epithelial ovarian cancer. Cancer 1989;63:1070-3. 4. Piver MS, Malfetano], Baker TR, Hempling RE. Five-year survival for stage Ie or stage I Grade 3 epithelial ovarian cancer treated with cisplatin-based chemotherapy. GynecolOncol 1992;46:357-60. 5. Creasman wr, DiSaia PJ. Screening in ovarian cancer. AM ] OBSTET GYNECOL 1991;165:7-10. 6. Westhoff C, Randall MC. Ovarian cancer screening: po-

24. 25.

26.

27.

28.

tential effect on mortality. AM] OBSTET GYNECOL 1991;165: 502-5. Granai CO. Ovarian cancer-unrealistic expectations. N Engl] Med 1992;327:197-9. Cramer DW, Hutchison GB, Welch WR, Scully RE, Ryan K]. Determinants of ovarian cancer risk. I. Reproductive experience and family history. ] Nat! Cancer Inst 1982; 71:711-6. Schildkraut]M, Thompson WD. Familial ovarian cancer: a population-based case-control study. Am ] Epidemiol 1988;128:456-66. Mori M, Harabuchi I, Miyake H, Casagrande ]T, Henderson BE, Ross RK. Reproductive, genetic, and dietary risk factors for ovarian cancer. Am ] Epidemiol 1988;128: 771-7. Lynch HT, Schuelke GS, Wells IC, et al. Hereditary ovarian carcinoma biomarker studies. Cancer 1985;55:410-5. Lynch HT, Fitzsimmons ML, Conway TA, Bewtra C, Lynch ]. Hereditary carcinoma of the ovary and associated cancers: a study of two families. Gynecol Oncol 1990;36: 48-55. Piver MS, Mett!in CJ, Tsukada Y, Nasca P, Greenwald P, McPhee ME. Familial ovarian cancer registry. Obstet Gynecol 1984;64: 195-9. Heintz APM, Hacker NF, Lagasse LD. Epidemiology and etiology of ovarian cancer: a review. Obstet Gynecol 1985; 66:127-35. Lynch HT, Harris RE, Guirgis HA, Maloney K, Carmody LL, Lynch]F. Familial association of breast/ovarian carcinoma. Cancer 1978;41:1543-9. Schildkraut ]M, Risch N, Thompson WD. Evaluating genetic association among ovarian, breast, and endometrial cancer: Evidence for a breast/ovarian cancer relationship. Am] Hum Genet 1989;45:521-9. Rodriguez MH, Platt LD, Medearis AL, Lacarra M, Lobo RA. The use of transvaginal sonography for evaluation of postmenopausal ovarian size and morphology. AM ] OBSTET GYNECOL 1988;159:810-4. van Nagell ]R, Higgins RV, Donaldson ES, et al. Transvaginal sonography as a screening method for ovarian cancer. Cancer 1990;65:573-7. van Nagell]R, DePriest PD, Puis LE, et al. Ovarian cancer screening in asymptomatic postmenopausal women by transvaginal sonography. Cancer 1991;68:458-62. Bourne TH, Whitehead MI, Campbell S, Royston P, Bhan V, Collins WP. Ultrasound screening for familial ovarian cancer. Gynecol Oncol 1991 ;43:92-7. Shimamoto K, Sakuma S, Ishigaki T, Makino N. Intratumoral blood flow: evaluation with color Doppler echography. Radiology 1987;165:683-5. Taylor K]W, Ramos I, Carter D, Morse SS, Snower D, Fortune K. Correlation of Doppler US tumor signals with neovascular morphologic features. Radiology 1988; 166: 57-62. Bourne T, Campbell S, Steer C, Whitehead MI, Collins WP. Transvaginal colour flow imaging: A possible new screening technique for ovarian cancer. Br Med] 1989; 299:1367-70. KUIjak A, Zalud J, Alfirevic Z. Evaluation of adnexal masses with transvaginal color ultrasound. ] Ultrasound Med 1991;10:295-7. Weiner Z, Thaler J, Beck D, Rotten S, Deutsch M, Brandes ]M. Differentiating malignant from benign ovarian tumors with transvaginal color flow imaging. Obstet Gynecol 1992;79:159-62. van der Burg MEL, Lammes FB, van Putten WL], Stoter G. Ovarian cancer: the prognostic value of the serum half-life of CA-125 during induction chemotherapy. GynecolOncol 1988;30:307-12. Jager W, Adam R, Wildt L, Lang S. Serum CA-125 as a guideline for the timing of a second-look operation and second-line treatment in ovarian cancer. Arch Gynecol Obstet 1988;243:91-9. Bast RC ]r, Siegal FP, Runowicz C, et al. Elevation of

Masse et al.

Volume 169, Number 3 Am J Obstet Gynecol

serum CA-125 prior to diagnosis of an epithelial ovarian carcinoma. Gynecol Oncol 1985;22:115-20. 29. Mann WJ, Patsner B, Cohen H, Loesch M. Pre-operative serum CA-125 levels in patients with surgical Stage I invasive ovarian adenocarcinoma. J Nat! Cancer Inst 1988; 80:208-9. 30. Einhorn N, Sjovall K, Knapp C, et al. Prospective evaluation of CA 125 levels for early detection of ovarian cancer. Obstet Gynecol 1992;80: 14-8. 31. Dnistrian AM, Schwartz MK. Plasma lipid-bound sialic acid and carcinoembryonic antigen in cancer patients. Clin Chern 1981;27:1737-9.

32. Knauf S, Anderson DJ, Knapp RC, Bast RC Jr. A study of the NB!70K and CA-125 monoclonal antibody radioimmunoassays for measuring serum antigen levels in ovarian cancer patients. AMJ OBSTET GYNECOL 1985;152:911-3. 33. Nam J-H, Cole LA, Chambers JT, Schwartz PE. Urinary gonadotropin fragment, a new tumor marker. I. Assay development and cancer specificity. Gynecol Oncol 1990; 36:383-90. 34. Slamon DJ, Godolphin W, Jones LA, et al. Studies of the HER-2/neu proto-oncogene in human breast and ovarian cancer. Science 1989;244:707-12.

A prospective study of several potential biologic markers for early prediction of the development of preeclampsia Jacques Masse, MD, MSc,· Jean-Claude Forest, MD, PhD: Jean-Marie Moutquin, MD,b Sylvie Marcoux, MD, PhD," Natalie-Anne Brideau, MSc, and Micheline Belanger, RN

Quebec, Quebec, Canada OBJECTIVE: The purpose of this study was to prospectively evaluate the predictive performance of several potential biologic markers of preeclampsia used alone or in combination. STUDY DESIGN: A prospective cohort of 1366 nulliparous women was followed up longitudinally on three occasions during pregnancy. The predictive performance of the tests, used either alone or in combination (stepwise multiple logistic regreSSion), was assessed and compared with that of the mean arterial pressure. RESULTS: Preeclampsia occurred in 109 of the pregnant women. At a specificity of 80% the sensitivity and the positive and negative predictive values for mean arterial pressure (at a threshold of 87 mm Hg) were 46.6%, 23.5%, and 92.0%, respectively, and the corresponding values for a multiple logistic model at 15 to 24 weeks that included some biologic markers, as well as the mean arterial pressure, were 57.1%,26.9%, and 93.7%, respectively. CONCLUSION: Preeclampsia can be predicted by a combination of simple biologic tests with a performance similar to second-trimester mean arterial pressure. However, this procedure is insufficient in terms of clinical usefulness. (AM J OBSTET GVNECOL 1993;169:501-8.)

Key words: Pregnancy complications, preeclampsia, calcium, microalbumin, progesterone Preeclampsia is a frequent cause of fetal or maternal morbidity and mortality. The cause is unknown, and delivery is still the only treatment once the disease has occurred. Recent studies have suggested that preeclampsia could be prevented, I but uncertainty about the innocuousness of the proposed interventions 2 still

From the Departments of Biochemistry, a Obstetrics and Gynecology, b and Social and Preventive Medicine,' Faculty ofMedicine, Universite Laval. Supported by a grant from the National Health Research and Development Program, Health and Welfare Canada (Project No. 6605-2988-52). Received for publication January 20, 1993; revised April 21, 1993; accepted May 6, 1993. Reprint requests: Jean-Claude Forest MD, PhD, Centre de recherche, Hopital Saint-Fran{ois d'Assise, Quebec, Quebec, Canada G1L3L5. Copyright © 1993 by Mosby-Year Book, 1nc. 0002-9378/93 $1.00 + .20 6/1/48453

precludes their widespread use on unselected populations. The selection of a high-risk population has been hindered by the lack of a predictive test for preeclampsia. Numerous tests, either clinical or biologic, have been reported, but their predictive performance remains debated or unproved." 4 Several flaws need to be avoided in the evaluation of the predictive performance of a test or a combination of tests for preeclampsia. The diagnosis is essentially clinical, and the frequency of erroneous classification is particularly high in multiparous women. The incidence of an elevated blood pressure reading for the first time during labor or in the early postpartum period is not associated with the increased complications found in cases of preeclampsia. 5 The timing of testing may affect the predictive power of the test and may invalidate its use as an early predictor of the disease if the result 501