The role of ultrasound evaluation in the detection of early-stage epithelial ovarian cancer

Transcription

1 American Journal of Obstetrics and Gynecology (2005) 192, The role of ultrasound evaluation in the detection of early-stage epithelial ovarian cancer David A. Fishman, MD, a, * Leeber Cohen, MD, b Stephanie V. Blank, MD, a Lee Shulman, MD, b Diljeet Singh, MD, MPH, b Kenny Bozorgi, MD, b Ralph Tamura, MD, b Ilan Timor-Tritsch, MD, a Peter E. Schwartz, MD c Section of Gynecologic Oncology, New York University School of Medicine, New York, NY, a Department of Obstetrics and Gynecology, Northwestern University, Evanston, Ill, b and Section of Gynecologic Oncology, Yale University School of Medicine, New Haven, Conn c KEY WORDS Epithelial ovarian cancer Ultrasound evaluation Early detection Objective: Epithelial ovarian cancer kills more women than all other gynecologic malignancies combined because of our inability to detect early-stage disease. Ultrasonography has demonstrated usefulness in the detection of ovarian cancer in asymptomatic women, but its value for the detection of early-stage epithelial ovarian cancer in women of increased risk is uncertain. We examined the usefulness of sonography in the detection of early-stage epithelial ovarian cancer in asymptomatic high-risk women who participated in the National Ovarian Cancer Early Detection Program. Study design: Only asymptomatic women of increased risk for the development of ovarian cancer with initial normal gynecologic and ultrasound examinations were eligible to participate in the institutional review board approved National Ovarian Cancer Early Detection Program. Participants underwent comprehensive gynecologic and ultrasound examinations every 6 months. Increased risk includes women with at least 1 affected first-degree relative with ovarian cancer; a personal history of breast, ovarian, or colon cancer; R1 affected first- and second-degree relatives with breast and or ovarian cancer; inheritance of a breast cancer mutation from an affected family member, or membership within a recognized cancer syndrome. Results: The average age of the 4526 women who were evaluated was 46 years; 2610 women were premenopausal, and 1916 women were postmenopausal. A total of 12,709 scans have been performed since Visualization of both ovaries was noted in 98% of premenopausal and in 94% of postmenopausal women. Fourteen women had undergone unilateral salpingooophorectomy. Recall rates at less than the routine 6-month interval were 0.4% in the premenopausal and 0.3% in postmenopausal women. A total of 98 women with persistent Supported by National Cancer Institute grant UO1CA85133; Early Detection Research Network; National Cancer Institute grant P50 CA83639; National Institutes of Health grants R01 CA89503, RO1 CA82562, and RO1 CA01015; Friends of Prentice Foundation; Northwestern Memorial Hospital; Stenn Fund for Ovarian Cancer Research; Joanne Silverman Cancer Foundation; Kaleidoscope of Hope Foundation; Illinois Department of Public Health; Lynne Cohen Foundation; Chicago Bears Care; Lynne Sage Foundation; the Robert H. Lurie Comprehensive Cancer Center, and the New York University Cancer Institute. Presented at the 23rd Annual Meeting of the American Gynecological and Obstetrical Society, September 9-11, 2004, Bolton Landing, NY. * Reprint requests: David A. Fishman, MD, Department of Obstetrics and Gynecology, New York University School of Medicine, 550 First Ave, NBV9N28, New York, NY David.Fishman@med.nyu.edu /$ - see front matter Ó 2005 Elsevier Inc. All rights reserved. doi: /j.ajog

2 Fishman et al 1215 adnexal masses were identified, and 49 invasive surgical procedures were performed that diagnosed 37 benign ovarian tumors and 12 gynecologic malignancies. All cancers were detected in asymptomatic women who had normal ultrasound and physical examinations 12 and 6 months before the cancer diagnosis. The detected malignancies were fallopian tube carcinoma (stage IIIC; n = 4 women), primary peritoneal carcinoma (n = 4 women; stage IIIA, 1 woman; stage IIIB, 2 women; stage IIIC, 1 woman), epithelial ovarian cancer (stages IIIA and IIIB; n = 2 women), and endometrial adenocarcinoma (stage IA; n = 2 women). Additionally 37 primary and 12 recurrent breast carcinomas were detected by physical examination. A total of 184 women with genetic predisposition (breast cancer positive) have undergone a prophylactic bilateral salpingooophorectomy; 23% of these procedures found atypical hyperplasia, and unexpectedly, 2 women (1%) were found to have stage III (A and B) primary peritoneal carcinoma. Conclusion: This study demonstrates the limited value of diagnostic ultrasound examination as an independent modality for the detection of early-stage epithelial ovarian cancer in asymptomatic women who are at increased risk for disease. Ó 2005 Elsevier Inc. All rights reserved. In the United States more women die annually from ovarian cancer than from all other gynecologic malignancies combined. This year it is anticipated that approximately 25,400 women will be diagnosed with ovarian cancer and that 14,300 women will die of this disease. 1 Unfortunately, 70% to 75% of women continue to be diagnosed with disseminated (stage III or IV) epithelial ovarian cancer (EOC) with a resultant overall 5-year survival rate of approximately 15%. This poor survival rate has remained essentially unchanged over the past 40 years, despite aggressive radical surgical intervention and new chemotherapies. However, if EOC is detected when the disease is confined to the ovary (stage I), the 5-year survival rate is approximately 90%, requires less radical surgical intervention, and may not require adjuvant chemotherapy. 2 Therefore, the ability to detect early-stage EOC accurately would have a significant impact on women s health care. The purpose of this study was to evaluate the contribution of ovarian ultrasound evaluation to the identification of early-stage epithelial ovarian malignancies. Material and methods Risk assessment The Yale University, Northwestern University, and New York University ovarian cancer early detection programs began in 1990, 1996, and 1999, respectively; in 1999 they became incorporated into the National Cancer Institute s Early Detection Research Network as the National Ovarian Cancer Early Detection Program (NOCEDP). All institutions use institutional review board approved protocols, and all participants provided written consent. Only asymptomatic women who were deemed to be at higher risk than that of the general population (1.8%) for the development of ovarian cancer with an initial normal gynecologic physical and ultrasound examination were eligible for participation in the early detection program. Women who were found to have an adnexal mass at initial examination were referred for appropriate further medical evaluation and were no longer eligible for participation within the NOCEDP. Increased risk for ovarian cancer was defined by the following criteria: (1) R1 affected first-degree relative with ovarian cancer; (2) a personal history of breast or ovarian cancer; (3) multiple first- and second-degree relatives with breast or ovarian cancer; (4) inheritance of a BRCA mutation from an affected family member; or (5) membership within a recognized family cancer syndrome (such as hereditary nonpolyposis colorectal carcinoma). Because the program was initiated in 1990, quantification of risk assessment has evolved, and presently multiple analytical programs are used (such Myriad, Salt Lake City, Utah, and BRCAPro, Dallas, Tex). On entry into the program, all women undergo a formal pedigree assessment by a board-certified expert in cancer genetics (genetic counselor and physician geneticist); those patients whose cases are deemed appropriate are offered genetic testing. Every 6 months participants undergo comprehensive physical examinations by board-certified gynecologic oncologists and ultrasound examination by sonologists with extensive experience (O15 years) in gynecologic ultrasonography and have investigational serum biomarker analysis. Additionally, all participants continue to have routine gynecologic care, which includes Papanicolaou tests and annual physical examinations by their referring health care providers. Ultrasound examination Transvaginal sonography was and remains the method of choice for ultrasound screening of the adnexa at all 3 institutions. Transabdominal sonography is used only in women whose adnexa image poorly or women who refuse a vaginal scan. Color Doppler evaluation was used routinely at the Yale program in the early 1990s.

3 1216 Fishman et al Figure 1 Simple ovarian cyst shows peripheral blood flow. Figure 2 Complex adnexal mass with multiple septations, yet without central blood flow that is consistent with a serous cystadenoma (confirmed by pathologic evaluation). For this study, color or power Doppler evaluation was only used as a secondary test and only when an adnexal mass was identified. Because ultrasound technology has improved rapidly, gynecologic evaluations were performed with a variety of equipment that included Acuson 128 XP, Acuson Sequoia, ATL 3000, 3500, and 5000 (Philips Corp, Bothell, Wash; Siemens Corp, Malvern, Pa). The evaluation process had adnexal masses graded as cystic, multiloculated, complex, or solid. An overall impression of malignancy risk was assigned to adnexal masses with a combination of gray-scale morphologic evidence and Doppler investigation, as we have previously described. 3 A variety of Doppler methods have been used that range from color Doppler evaluation with velocimetry in the early 1990s to 3-dimensional power Doppler mapping of blood flow patterns starting in This latter technique, which is being used presently, evaluates masses for the presence of peripheral as compared with central vascular blood flow, which we have found to improve our specificity significantly to discern benign from malignant tumors (Figures 1-3). 4 Premenopausal women with simple cysts or hemorrhagiclike cysts are rescanned routinely in 6 to 8 weeks to check for resolution. Menopausal women with stable, simple, thin-walled cystic masses that are!5 cm with no evidence of central blood flow are rescanned in 2 months and then at the routine 6-month intervals to assure stability. Patients with suspected hydrosalpinges or pelvic adhesive disease are treated similarly. Results A total of 12,709 scans have been performed on 4526 women (premenopausal, 2610; postmenopausal, 1916) Figure 3 Complex adnexal mass with multiple septations with central blood flow that is consistent with a serous cystadenocarcinoma (confirmed FIGO stage IIIC). since Visualization of both ovaries was noted in 98% of premenopausal women and 94% of menopausal women. Recall rates at less than the routine 6-month interval were 0.4% in the premenopausal group and 0.3 % in menopausal women. Cystic masses O2.5 cm. were visualized in 12.6% of premenopausal women and 6.4% of menopausal women. A total of 98 persistent adnexal masses were identified, and 49 operations were performed for benign and malignant disease. All women had been followed for at least 12 months before the diagnosis of an abnormal adnexal mass and have been examined by their referring physician before surgical intervention. Operative intervention yielded 12 gynecologic malignancies and 37 benign ovarian tumors or 3.1 benign masses per gynecologic malignancy. Adnexal masses were evaluated with 3-dimensional

4 Fishman et al 1217 Table I Case Cancer site Gynecologic malignancies Histologic condition Stage/ grade Age at diagnosis (y) Mutation status Visits (n) Ultrasound findings at diagnosis 1 Primary peritoneal Papillary serous IIIC/3 52 ÿ 4 Normal ovaries Deceased Primary peritoneal Papillary serous IIIB/3 38 C 6 Normal ovaries Alive 45 3 Primary peritoneal Papillary serous IIIA/3 66 C 4 Normal ovaries Deceased 86 4 Fallopian tube Papillary IIIC/3 56 C 6 Dilated fallopian tube; normal ovaries Deceased Fallopian tube Papillary serous-g3 IIIC 64 ÿ 8 Complex adnexal mass Alive Fallopian tube Papillary IIIC/3 43 ÿ 4 Dilated fallopian tube; Alive 69 normal ovaries 7 Fallopian tube Papillary IIIC 58 ÿ 2 Dilated fallopian tube Deceased 31 8 Ovary Papillary serous IIIC 62 ÿ 7 Complex adnexal mass Deceased 45 9 Ovary Papillary serous IIIC 60 ÿ 6 Complex adnexal mass Alive Endometrial Endometrioid 1A/1 64 ÿ 6 Normal ovaries NED 11 Endometrial Endometrioid 1A/1 66 ÿ 3 Normal ovaries; abnormal uterine stripe Alive 12 Primary peritoneal Papillary serous IIIB 49 ÿ 4 Normal ovaries Alive 89 Status Ca125 value at diagnosis (U/mL) ultrasonography (Figures 1-3), and we previously have reported that the location of vascular flow (central vs peripheral) is critical in the determination of malignancy. 3 Figures 1 through 3 show the architectural and vascular measures that were used to help identify those masses that were suspicious of malignancy. Table I shows the malignancies that were detected within the program: 4 primary peritoneal carcinomas (1 stage IIIA, 2 stage IIIB, 1 stage IIIC), 4 fallopian tube carcinomas (3 stage IIIC, 1 stage IV), 2 EOCs (stage IIIC), and 2 endometrial carcinoma (stage IA). Additionally, 184 women with an inherited BRCA-1 or -2 mutation had prophylactic bilateral salpingooophorectomy; 23% of these women had evidence of atypical ovarian or fallopian hyperplasia. Two asymptomatic BRCAC women unexpectedly were found to have primary peritoneal carcinoma (1 stage IIIA, 1 stage IIIB) at diagnostic laparoscopy for a scheduled prophylactic bilateral salpingo-oophorectomy (2/184; 1%; Table II). Of the women with gynecologic malignancies, excluding endometrial cancer, 4 women (40%) were BRCA 1C (2 primary peritoneal carcinomas, 2 fallopian tube carcinoma). Interestingly, 37 participants were found to have an early-stage BRCA that was identified on physical examination, of which 6 of the malignancies were BRCA 1 or 2 C (22%). Comment The NOCEDP as part of the National Cancer Institute s Early Detection Research Network is committed to the development of effective means for the accurate detection of early-stage EOC. Within the NOCEDP the identification of clinically relevant serum/plasma markers that are based on ovarian carcinogenesis, Table II Number of BRCAC women who received prophylactic surgery BRCAC, prophylactic bilateral salpingo-oophorectomy (n) Pathologic finding 162 (Total) Benign Stage IIIA and B primary peritoneal carcinoma (n = 2) invasion, and metastasis are combined with newly developed diagnostic technologies to achieve this goal. Despite the significant advances in diagnostic sonography, we report its present inability as an independent modality to detect early-stage EOC in asymptomatic women who are deemed to be at increased risk for the development of this disease. In 1991 Bourne et al 3 suggested that transvaginal ultrasound evaluation could provide effective screening for early-stage ovarian cancer in high-risk women because such evaluation detected 2 stage IA ovarian cancers and 3 stage I borderline tumors in 776 women. In 1998 Bell et al 5 reviewed the literature and identified 6 studies with a combined total of 3146 high-risk women on the basis of family history. Excluding borderline tumors, 8 ovarian cancers were identified, of which 2 cancers were stage I, with a prevalence rate of 254 of every 100,000 tumors, which is approximately 6 times the expected prevalence rate of 40 of every 100,000 tumors in low risk women. Bell et al calculated that 25% of the ovarian cancers were diagnosed at stage I, with a 95% CI of 3% to 65%. We were able to identify asymptomatic advancedstage adnexal malignancies; yet, not 1 early-stage EOC

5 1218 Fishman et al was diagnosed in our study population. The inability of diagnostic ultrasound evaluation per se to identify stage I ovarian cancer routinely in high-risk women has also been reported by Karlan et al, 6 who diagnosed 1 stage I epithelial ovarian cancer and 7 stage III primary peritoneal cancers in the 1261 women who participated in their program. 7 Similarly, the 4 women with stage III primary peritoneal cancer within our population had normal physical and ultrasound examinations. Clearly, ultrasound examination would not be expected to be useful to screen for early-stage primary peritoneal cancer because adnexal disease defines widespread dissemination. It is possible that normal low-risk screening may identify a higher proportion of women with a diagnosis of stage I EOC. In recent reports, Van Nagell et al 8 discussed their 12-year study that evaluated 14,469 women and found that transvaginal ultrasound examination detected 52% of the ovarian cancer cases at stage I (11/21 cases). However, if granulosa cell and borderline tumors are excluded, the detection rate decreased to 31% (5/16) cases. 9 In the study of Van Nagell et al, they included high-risk women O25 years of age and all women, regardless of risk, who were O50 years of age. In the previously mentioned analysis by Bell et al, 5 ultrasound screening studies with 15,834 low-risk women were summarized. The proportion of women who were diagnosed at stage I, excluding borderline tumors, was 4:6 (67%; 95% CI, 22%-96%). Compared with our data, this suggests that a higher percentage of stage I ovarian cancers may be detected in low-risk women than in our high-risk population. These studies also significantly differed from ours on the basis of inclusion and exclusion criteria because we excluded those women with adnexal abnormalities that were detected on the initial examination. The British National Health Service is now conducting a prospective trial of 200,000 menopausal women, using ultrasound architectural analysis without vascular evaluation as a primary screen in 50,000 women and serum Ca125 determinations as a primary screen in 50,000 women, with 100,000 women serving as control subjects. 10 The work of Jacobs et al 11,12 has demonstrated the power of multimodal screening. A disease with a low annual prevalence requires a highly specific initial screening test. After menopause, serum Ca125 has a false-positive rate of only 1%. Ca125 in combination with ultrasound evaluation, as demonstrated by Jacobs et al, has a 2% false-positive rate. Unfortunately, among 22,000 menopausal low-risk women, the sensitivity for stage I disease was only 16% (3/19 women) for stage I disease. Furthermore, Einhorn et al 13 have demonstrated that screening premenopausal women markedly increases the false-positive screen rate to 10%. Therefore, a more sensitive biologic marker would be useful. We identified and clinically followed 98 persistent complex adnexal masses in our patient population, which ultimately led to 49 operations, 12 operations for gynecologic malignancies and 37 operations for benign adnexal tumors. These adnexal masses developed in women who on initial evaluation were found to have normal gynecologic and ultrasound examination results and were also normal for at least 12 months before the diagnosis. For those 12 women who were found to have gynecologic malignancies, there was a minimum of 3 normal consecutive examinations before the diagnosis of an adnexal abnormality. The operative intervention rate in our patient population, which included 2 endometrial adenocarcinomas, was 3.1 benign adnexal masses for each gynecologic malignancy, which is at the low end for reported programs. A review of prospective ultrasound studies as a primary screen reveals surgical intervention rates that range from 3 to 50 operations for each ovarian cancer that was detected. 3,5,14 The lowest rates were seen in programs that combined transvaginal gray-scale analysis with Doppler evaluation as a secondary tool. The highest reported rate was the use of only an abdominal scan without Doppler evaluation. A recent meta-analysis has shown that the combination of grayscale ultrasound evaluation and Doppler imaging performs more accurately than either modality alone. 15 There is no evidence for the initial promise that stage I EOC might be identified in ovaries of normal size and appearance with Doppler technology. 16 The role of Doppler velocimetry in distinguishing benign from malignant ovarian disease has been controversial since the early 1990s. In 1993, Fleischer and Jones 17 argued for a multiparameter approach that involved both gray- scale and Doppler vascular architecture. They noted that malignant tumors frequently display low-resistance vessels but warned against using arbitrary resistance cut-offs. The controversy over Doppler evaluation was also reviewed by Karlan and Platt 18 in The National Institute of Health in 1995 stated that Doppler evaluation might improve specificity, but its use was investigational. 19 Logistic regression analyses by Tailor et al, 20 Timmerman et al, 21 Schelling et al, 22 and Alcazar and Jurado 23 have all shown that the most important variable in predicting malignancy is the presence or absence of solid elements within the tumor. It remains controversial whether vascular blood flow should be scored with velocimetry measurements or by a subjective grading of the overall vessel angioarchitecture of the mass. Unfortunately, papers by Aslam et al 24 and Valentin et al 25 have not confirmed that any logistic regression formulae perform as well as when initially published. We, and others, have previously reported that overall angioarchitecture may be better seen with power Doppler imaging than with spectral color Doppler imaging. 3,4,20,26-29 Power Doppler imaging is less angle

6 Fishman et al 1219 dependent and displays low-velocity flow vessels better but is subject to motion artifact. We found that specificity significantly improved from 54% to 75% by the addition of 3-dimensional Doppler imaging as a secondary test to determine the location of blood flow (central vs peripheral). 4 Guerriero et al 29,30 argued that there is little evidence that 3-dimensional Doppler imaging is significantly better than results with 2-dimensional gray-scale and color Doppler imaging. Tailor et al 20 recently presented the results of their ultrasound-based familial ovarian cancer program that evaluated 2500 asymptomatic women with at least one close relative with ovarian cancer. A total of 4231 screens were performed, of which 2500 were initial screens, 998 were second scans, and 733 were further repeat scans. They reported that their screening program identified 11 EOCs, of which 7 cancers (64%) were stage I disease. However, 4 of these early-stage ovarian cancers were classified inappropriately, because they were borderline ovarian tumors; all 3 stage I EOCs were detected at the initial screening ultrasound examination. One of the inclusion criteria to enter our NOCEDP is that the initial sonographic examination must be normal because any woman with an ovarian mass is ineligible for participation. Therefore, following our inclusion/ exclusion criteria, their screening program would not have identified any stage I EOCs, which has been our experience. The NOCEDP clinical experience is similar to that reported by Karlan et al, 6 in that ultrasound evaluation as an independent modality has poor sensitivity for the detection of stage I EOC in the high-risk population. Despite intensive and expert sonographic surveillance, not one early-stage adnexal malignancy was identified, and more women were diagnosed with breast, primary peritoneal, and fallopian cancer than ovarian cancer. Our study again questions whether women within the high-risk population are more likely to have primary peritoneal and fallopian carcinomas and ovarian malignancies. Based on the recent literature discussed earlier, it is possible that screening ultrasound evaluation is more sensitive in the low-risk population. The ability to detect early-stage EOCs by a simple test has long been an objective in medical screening. A limitation of the currently available individual tests is the lack of sensitivity and specificity when applied to large heterogeneous populations. Our NOCEDP was established specifically to apply the biochemical, genetic, and molecular knowledge of ovarian carcinogenesis, invasion, and metastasis clinically to address the problem of detection of early rather than late stage EOC. The enhanced understanding of ovarian cancer biologic condition has led to the identification and detection of specific genetic, molecular, and serum biomarkers in women with ovarian cancer that may have clinical usefulness in the evaluation of women who are deemed at increased risk for the development of this disease. Serum and plasma markers (such as lysophosphatidic acid, autotaxin, soluble epidermal growth factor receptors, matrix metalloproteinases, and kallikreins) all appear to have clinical usefulness Previously, we demonstrated that low molecular weight serum proteomic patterns from low-resolution surface-enhanced laser desorption/ionization time-of-flight mass spectrometer data could distinguish neoplastic from non-neoplastic disease within the ovary. 37 Using newly developed highresolution mass spectrometer technology, we recently reported that the use of serum proteomic pattern analysis for the generation of multiple highly accurate models that were obtained with a hybrid quadruple time-of-flight mass spectrometer achieved 100% sensitivity and specificity for the diagnosis of early-stage EOC. 38 Initial results with proteomics and other newly developed biomarker assays appear clinically promising and suggest high specificity and sensitivity for the detection of EOC and early-stage disease. 37 Our recent reports regarding proteomics suggest that screening for ovarian cancer ultimately may follow a similar model as seen for prenatal Down syndrome screening (ie, first 1 marker [maternal age] followed by both multiple serum markers [eg, human chorionic gonadotropin, estriol, inhibin, PAPP-A], and biophysical markers on ultrasound evaluation [ie, nuchal translucency, short femur length]). The clinical usefulness of such serum/plasma markers independently or in combination is currently under investigation by multi-institutional, multinational clinical trials within the NOCEDP. Continued technologic advances in diagnostic imaging (such as contrast enhanced imaging) may improve the architectural and vascular visualization that are afforded by abdominal and transvaginal ultrasound evaluation significantly, thereby making it possible to discern subtle microvascular and metabolic changes within the tumor microenvironment before the presence of a discernable mass. In summary, our ongoing study demonstrates the limited value of diagnostic ultrasound evaluation as an independent primary screening tool for the detection of earlystage EOC in asymptomatic women who are at increased risk. It is anticipated that the combination of the newly developed serum biomarkers in combination with high-resolution ultrasound evaluation will improve women s health care by accurately detecting early-stage EOC. Acknowledgments We thank the late Dr Kenneth Taylor for his participation, support, and guidance in the development of the NOCEDP and in the preparation of this manuscript.

7 1220 Fishman et al References 1. Jemal A, Murray T, Samuels A, Ghafoor A, Ward E, Thun MJ, et al. Cancer statistics, CA Cancer J Clin 2003;53: Yancik R. Ovarian cancer. age contrasts in incidence, histology, disease stage at diagnosis, and mortality. Cancer 1993;71: Cohen LS, Escobar PF, Scharm C, Glimco B, Fishman DA. Three-dimensional power Doppler improves the diagnostic accuracy for ovarian cancer prediction. Gynecol Oncol 2001; 82: Bourne TH, Campbell S, Reynolds KM, Whitehead MI, Hampson J, Royston P, et al. Screening for early familial ovarian cancer with transvaginal ultrasonography and color blood low imaging. BMJ 1993;306: Bell R, Petticrew M, Sheldon T. The performance of screening tests for ovarian cancer: results of a systematic review. BJOG 1998;105: Karlan BY, Baldwin RL, Lopez-Luevanos E, Raffel LJ, Barbuto D, Narod S, et al. Peritoneal serous papillary carcinoma, a phenotypic variant of familial ovarian cancer: implications for ovarian cancer screening. Am J Obstet Gynecol 1999;180: Liede A, Karlan BY, Baldwin RL, Platt LD, Kuperstein G, Narod SA. Cancer incidence in a population of Jewish women at risk for ovarian cancer. J Clin Oncol 2002;20: van Nagell JR Jr, DePriest PD, Reedy MB, Gallion HH, Ueland FR, Pavlik EJ, et al. The efficacy of transvaginal sonographic screening in asymptomatic women at risk for ovarian cancer. Gynecol Oncol 2000;77: Fishman D, Cohen L. Is transvaginal ultrasound effective for screening asymptomatic women for the detection of early stage epithelial ovarian carcinoma? Gynecol Oncol 2000;77: Jacobs IJ, Skates SJ, MacDonald N, Menon U, Rosenthal AN, Davies AP, et al. Screening for ovarian cancer: a pilot randomised controlled trial. Lancet 1999;353: Jacobs I, Stabile I, Bridges J, Kemsley P, Reynolds C, Grudzinskas J, et al. Multimodal approach to screening for ovarian cancer. Lancet 1988;1: Jacobs I, Davies AP, Bridges J, Stabile I, Fay T, Lower A, et al. Prevalence screening for ovarian cancer in postmenopausal women by CA 125 measurement and ultrasonography. BMJ 1993; 306: Einhorn N, Sjovall K, Knapp RC, Hall P, Scully RE, Bast RC Jr, et al. Prospective evaluation of serum CA125 levels for early detection of ovarian cancer. Obstet Gynecol 1992;80: Campbell S, Bhan V, Royston P, Whitehead MI, Collins WP. Transabdominal screening for early ovarian cancer. BMJ 1989; 299: Kinkel K, Hricak H, Lu Y, Kyo T, Filly RA. US characterization of ovarian masses: a meta-analysis. Radiology 2000;217: Kurjak A, Zalud I, Alfirevic Z. Evaluation of adnexal masses with transvaginal color ultrasound. J Ultrasound Med 1991; 10: Fleischer A, Jones H III. Color Doppler sonography of ovarian masses: the importance of a multiparameter approach [editorial]. Gynecol Oncol 1993;50: Karlan BY, Platt LD. The current status of ultrasound and color Doppler in imaging in screening for ovarian cancer. Gynecol Oncol 1994;55:S Consensus Development Panel. Ovarian cancer. JAMA 1995; 273: Tailor A, Jurkovic D, Bourne T. Sonographic prediction of malignancy in adnexal masses using multivariate logistic regression analysis. Ultrasound Obstet Gynecol 1997;10: Timmerman D, Bourne TH, Tailor A, Collins WP, Verrelst H, Vandenberghe K, et al. A comparison of methods for preoperative discrimination between malignant and benign masses: the development of a new logistic regression model. Am J Obstet Gynecol 1999;181: Schelling M, Braun M, Kuhn W, Bogner G, Gruber R, Gnirs J, et al. Combined transvaginal B-mode and color Doppler sonography for differential diagnosis of ovarian tumors: results of a multivariate logistic regression analysis. Gynecol Oncol 2000;77: Alcazar J, Jurado M. Prospective evaluation of a logistic model based on sonographic morphologic and color Doppler findings developed to predict adnexal malignancy. J Ultrasound Med 1999;18: Aslam N, Banerjee S, Carr JV, Savvas M, Hooper R, Jurkovic D. Prospective evaluation of logistic regression models for the diagnosis of ovarian cancer. Obstet Gynecol 2000;96: Valentin L, Hagen B, Tingulstad S, Eik-Nes S. Comparison of pattern recognition and logistic regression models for discrimination between benign pelvic masses: a prospective crossvalidation. Ultrasound Obstet Gynecol 2001;18: Guerriero S, Ajossa S, Risalvato A, Lai MP, Mais V, Angiolucci M, et al. Diagnosis of adnexal malignancies by using color Doppler energy imaging as a secondary test in persistent masses. Ultrasound Obstet Gynecol 1998;11: Kurjak A, Kupesic S, Breyer B. The assessment of ovarian tumor angiogenesis: What does three-dimensional power Doppler add? Ultrasound Obstet Gynecol 1998;12: Kurjak A, Kupesic S, Tomislav A, Kosuta D. Three-dimensional ultrasound and power Doppler improve the diagnosis of ovarian lesions. Gynecol Oncol 2000;76: Guerriero S, Alcazar JL, Ajossa S, Lai MP, Errasti T, Mallarini G, et al. Comparison of conventional color and power Doppler imaging for the diagnosis of ovarian cancer: results of a European study. Gynecol Oncol 2001;83: Guerriero S, Alcazar JL, Coccia ME, Ajossa S, Scarselli G, Boi M, et al. Complex pelvic mass as a target of evaluation of vessel distribution by color Doppler sonography for the diagnosis of adnexal malignancies: results of a Multicenter European study. J Ultrasound Med 2002;21: Fishman DA, Singh D, Lurain JR, Bozorgi K. The scientific basis for the early detection of early stage epithelial ovarian carcinoma: part 1. The Female Patient 2002;27: Fishman DA, Singh D, Lurain JR, Bozorgi K. The scientific basis for the early detection of early stage epithelial ovarian carcinoma: part 2, the future of early detection. The Female Patient 2002;27: Baron AT, Cora EM, Lafky JM, Boardman CH, Buenafe MC, Rademaker A, et al. Soluble epidermal growth factor receptor (segfr/serbb1) as a potential risk, screening, and diagnostic serum biomarker of epithelial ovarian cancer. Cancer Epi Bio Prev 2003;12: Baron AT, Cora EM, Lafky JM, Boardman CH, Buenafe MC, Rademaker A, et al. Serum serbb1 and epidermal growth factor levels as tumor biomarkers in women with stage III or IV epithelial ovarian cancer. Cancer Epi Bio Prev 1999;8: Mills GB, Eder A, Fang X, Hasegawa Y, Mao M, Lu Y, et al. Critical role of lysophospholipids in the pathophysiology, diagnosis, and management of ovarian cancer. Cancer Treat Res 2002;107: Tanyi JL, Morris AJ, Wolf JK, Bast RC, Lu K, Smith D, et al. Role of decreased levels of LPP-1 in accumulation of lysophosphatidic acid (LPA) in ovarian cancer. Clin Cancer Res 2003; 9: Petricoin EF III, Ardekani A, Hitt BA, Levine PJ, Fusaro VA, Steinberg SM, et al. Serum proteomic patterns discovered using an artificial intelligence based algorithm reveal diseaseassociated signatures of ovarian and prostate cancer. Lancet 2002; 359:572-7.

8 Fishman et al Conrads TP, Fusaro VA, Ross S, Hitt BA, Chen TH, Levine PJ, et al. Multiple high-resolution serum proteomic patterns for ovarian cancer detection. Enocrine-Related Cancer 2004;11: Discussion DR J. R. VAN NAGELL, JR, Lexington, Ky. Dr Fishman et al have presented data from a screening study that was conducted at Northwestern and Yale Universities on a population of 4526 asymptomatic women who were at high risk for ovarian cancer from 1990 to the present. Transvaginal sonography and biomarker analysis were performed at 6-month intervals on these patients. Ninety-six patients were identified as having a persistent adnexal mass, and 48 patients underwent operation. Eleven of these patients had gynecologic malignancies that included 3 primary peritoneal carcinomas (1 stage IIIA, 1 stage IIIB, 1 stage IIIC), 4 fallopian tube carcinomas (3 stage IIIC, 1 stage IV), and 2 EOCs (stage IIIC). On the basis of this data, they conclude that sonography as an independent modality is unable to detect early-stage EOC in a patient population that is at high risk for the disease. Because sonographic detection of ovarian cancer is based on the observation of an abnormality in ovarian volume or morphologic condition, it is logical that transvaginal sonography would not detect reliably either early-stage primary peritoneal cancer or early-stage fallopian tube cancer. Both of these malignancies at an early stage would not produce abnormalities in ovarian volume or morphologic condition. Nevertheless, EOCs more frequently are associated with changes in ovarian size and structure, which often can be detected sonographically. In the University of Kentucky Ovarian Cancer Screening Project, we have found, in a lower risk population of >20,000 women whose cases have been followed for a mean of 5 years, that 45% of EOCs were detected at stage I provided that the women were screened regularly at 12-month intervals and did not miss a scheduled screening visit. 1 Likewise, Tailor et al 2 reported that transvaginal sonographic screening at 6- to 12-month intervals detected 7 invasive EOCs and 4 borderline ovarian epithelial tumors in 2400 asymptomatic women with at least 1 close relative who had ovarian cancer. Three of the 7 patients (43%) with invasive ovarian cancer were detected with stage I disease. Although this study presents data on 12,989 scans, one would expect a much higher number of scans because 4526 women were studied at 6-month intervals beginning in In this regard, please describe the exact screening algorithm that was used. How many patients who were enrolled in the trial missed a scheduled screening visit or withdrew from the study? The authors state that women who were found to have an adnexal mass on initial examination were referred for further evaluation and were excluded from the screening study. How many women on initial sonographic evaluation had ovarian or adnexal tumors detected, and how many of these cases were ovarian malignancies? This is of interest because 42% of all cases of ovarian cancer that were detected in the University of Kentucky Ovarian Cancer Screening Project were diagnosed on the first screen. 1 In our screening algorithm, postmenopausal patients with complex ovarian tumors (>10 cm 3 volume) that contain solid areas or wall papillations that persist on a repeat sonogram in 4 to 6 weeks are advised to have surgical tumor removal, preferably by laparoscopy. In this trial, 58% of patients were premenopausal. Ninety-six patients had persisting adnexal masses, but only 48 patients underwent operation. What morphologic, Doppler, or volume criteria were used to determine which patients had surgery and which were followed? Both cases of EOC that were reported in this series had a number of screening visits before detection. What were the ovarian tumor dimensions in these 2 patients at the time of sonographic diagnosis of ovarian cancer? What were the ovarian dimensions in these same 2 patients 6 months earlier? Also, both patients with ovarian cancer had elevated serum Ca125 values at the time of detection (45 U/mL and 64 U/mL, respectively). What were the serum Ca125 values in these 2 patients 6 months earlier? Finally, were there any documented cases of ovarian cancer in this series that were missed entirely by sonography (false-negative cases). If not, then I believe it may be premature to make definite conclusions about the efficacy of sonography in the early detection of EOC on the basis of the data from the relatively small sample size presented herein. Information generated from screening trials has increased our knowledge significantly concerning the biologic condition of ovarian cancer. The design of an optimal algorithm for the detection of early-stage ovarian cancer remains a challenge because of the relatively low prevalence of the disease, the variation in individual tumor growth rates, and the lack of sensitivity, specificity, and positive predictive value of the individual screening tests that are available. Hopefully, a combination of advanced sonography, novel biomarkers, and proteomic analysis will enable us to detect ovarian cancer at a stage when it is curable. References 1. van Nagell JR Jr, DePriest PD, Reedy MB, Gallion HH, Ueland FR, Pavlik EJ, et al. The efficacy of transvaginal sonographic