Screening for ovarian cancer

Author
Karen J Carlson, MD Section Editors
Barbara Goff, MD
Suzanne W Fletcher, MD Deputy Editor
H Nancy Sokol, MD

Last literature review version 17.1: January 2009 | This topic last updated: October 6,
2008 (More)

INTRODUCTION Ovarian cancer is the leading cause of death from gynecologic malignancy in
the United States. Approximately 20,180 cases are diagnosed annually, and there are 15,310
deaths attributable to ovarian cancer each year [1] . The lifetime probability of developing
ovarian cancer is 1.8 percent. The incidence of ovarian cancer increases with age; the highest
proportion of cases are diagnosed in women 50 to 59 years of age.

Interest in early detection as a method of reducing mortality has grown with the discovery of
tumor markers associated with ovarian malignancies (particularly CA 125) and with improved
diagnostic accuracy of pelvic ultrasonography. Large scale prospective clinical screening trials
are in progress to determine whether either approach reduces mortality from ovarian cancer.
Until the results of these trials are final, there is a consensus that women at average risk for
ovarian cancer should not undergo screening with either method outside these clinical trials.
Intensive research is in progress to identify additional markers and a cost-effective screening
strategy for this population of women.

In comparison, women with a family history of ovarian cancer are at higher risk of the disease.
Evidence suggests screening is appropriate for some of these women. Differentiation between
women with a suspected hereditary cancer syndrome and all others with a family history guides
screening recommendations for this group.

This topic will review the risks and benefits of screening for ovarian cancer in asymptomatic
women. The issue of testing for ovarian cancer in women with nonspecific symptoms that may

be associated with ovarian cancer is discussed separately. (See "Early detection of ovarian
cancer: role of symptom recognition").

BIOLOGICAL BASIS FOR SCREENING Survival from ovarian cancer is related to the stage at
diagnosis; five-year survival is over 90 percent for the minority of women with stage I disease [2]
. This number drops to about 75 to 80 percent with regional disease, and 25 percent for those
with distant metastases.

Despite the good prognosis associated with early stage disease, overall five-year survival in
women with ovarian cancer is less than 45 percent, in large part because the cancer has spread
beyond the ovary at the time of clinical detection in 75 percent of patients. Mortality from
ovarian cancer has decreased only slightly in the past 30 years [2] . These observations, coupled
with new developments in diagnostic methods, have fueled interest in early detection as an
approach to reducing mortality from this disease.

Little is known about the mechanism or timing of progression from localized to disseminated
ovarian cancer. The model of unifocal disease progressing to diffuse disease is plausible; ovarian
cancer also may develop from multiple foci within the abdomen since carcinomatosis can
develop after the removal of normal ovaries [3] .

Ninety percent of ovarian malignancies are epithelial tumors (including serous, mucinous,
endometrioid, and clear cell tumors); the remainder arise from germ cell and other tissues. A
subgroup of epithelial tumors, known as "borderline tumors" or "cystadenomas of low malignant
potential," have a much more favorable prognosis. Benign cystadenomas, another subgroup of
epithelial tumors, have a very low rate of malignant transformation. The benefit of identifying
asymptomatic benign cystadenomas is unknown.

RISK FACTORS Identification of women at high risk for ovarian cancer may help to select a
group that would most benefit from screening strategies. A number of risk factors for ovarian
cancer have been identified (show table 1). (See "Epithelial ovarian cancer: Pathogenesis,
epidemiology, and risk factors").

Genetic predisposition The strongest known risk factor for ovarian cancer is a family history,
which is present in about 10 to 15 percent of women who develop the disease [4] . When a
woman reports that one or more family members have had ovarian cancer, it is important to
differentiate the rare "familial ovarian cancer syndromes" from the more common "family history
of ovarian cancer;" the latter describes families that include isolated members with ovarian
cancer, without evidence of a hereditary pattern.

The risk of ovarian cancer is modestly increased when the family history suggests a sporadic
case rather than a hereditary cancer syndrome. A meta-analysis of pooled case-control studies
calculated an odds ratio of 3.1 for developing ovarian cancer in women with one first or seconddegree relative with the disorder [4] . Based upon these data, it was estimated that a family
history of ovarian cancer in one relative increased the lifetime probability of ovarian cancer in a
35-year-old woman from 1.6 to 5 percent. In contrast, hereditary syndromes are associated with
a lifetime risk of ovarian cancer as high as 50 percent. With wider availability of genetic testing,
identification of women with possible hereditary cancer syndromes has become increasingly
important.

Familial ovarian cancer syndromes Familial ovarian cancer syndromes are uncommon,
accounting for between 5 and 10 percent of ovarian cancer cases in most series [5-7] . These
syndromes include: Breast-ovarian cancer syndrome The Lynch II syndrome, involving cancers of
the colon, breast, endometrium, and ovary with hereditary nonpolyposis colorectal cancer
(HNPCC) Site-specific ovarian cancer syndrome

The most common hereditary syndrome is the breast-ovarian cancer syndrome. Most of these
families have germ-line mutations in one of the breast cancer susceptibility genes, BRCA1 or
BRCA2. (See "Genetic testing for breast and ovarian cancer"). In the United States, carriers of
BRCA mutations are rare in the general population (1 in 300 or fewer), but among persons of
Ashkenazi Jewish descent the prevalence is estimated to be 2 percent [8] .

Clues to the presence of a hereditary cancer syndrome include the presentation of ovarian
cancer in a family member at an early age (under 50 years) and the occurrence of ovarian or
related cancers in multiple members (ie, two to four generations) (show table 2). In women of
Ashkenazi Jewish descent (comprising the great majority of Jewish women in the United States),
some experts suggest the threshold for suspecting a familial cancer should be lower, with
consideration of genetic testing, if there is a family member with breast cancer before age 50 or
with ovarian cancer [8] .

The absolute risk of developing ovarian cancer over a lifetime associated with the presence of a
BRCA1 mutation is 40 to 50 percent, while it is less for those with BRCA2 mutations (15 to 25
percent) (show table 3). Population-based series [9-12] tend to report lower risks than those that
study women who are followed in high-risk hereditary cancer clinics [13] . This topic is addressed
in detail elsewhere. (See "Risk assessment and clinical characteristics of women with a family
history of breast and/or ovarian cancer", section on Ovarian cancer).

BRCA mutations may not be limited to women with a strong family history. One study of 116
unselected women with ovarian cancer identified 10 women with BRCA1 mutations and a single
BRCA2 mutation, which occurred in a patient also carrying a BRCA1 mutation [14] . Over one-half
of the women with BRCA1 mutations had unremarkable family histories. In addition, the majority
of patients who had a family history that was suggestive of a hereditary cancer syndrome

(maternal family history of breast or ovarian cancer) tested negative for BRCA gene mutations.
BRCA dysfunction, distinct from the recognized germline mutations, also appears to be
associated with ovarian cancer [15,16] ; however, the utility of testing for BRCA dysfunction has
not been proven, and no such test is clinically available.

Age The incidence of ovarian cancer increases with age. In women 50 to 75 years of age, the
annual incidence is 50 per 100,000 (adjusted for prior oophorectomy), approximately twice the
rate in younger women [17] . The likelihood that a case of ovarian cancer is attributable to a
gene mutation decreases with increasing age at diagnosis [18] .

Reproductive factors Several potentially modifiable reproductive factors appear to reduce the
risk of ovarian cancer (show table 1): Pregnancy reduces the odds of ovarian cancer by 25 to 50
percent [19] . The decrease in risk is associated with an increasing number of pregnancies. Use
of the oral contraceptive pill is associated with a 40 percent reduction in risk of ovarian cancer;
increasing duration of use is associated with decreasing risk [20] . One analysis estimated that
after five years of oral contraceptive use, nulliparous women can reduce their ovarian cancer risk
to the level seen in parous women who never used oral contraceptives; 10 years of use by
women with a positive family history can reduce their risk to a level below that for women with
no family history who never used oral contraceptives [21] .

Data on the impact of oral contraceptive use on the risk of hereditary forms of ovarian cancer,
such as those associated with mutations in the BRCA1 or BRCA2 gene, are conflicting. While one
case-control study reported a significant reduction in risk [22] , a second did not [23] .
Breastfeeding is associated with a more modest effect on risk, reducing the odds of ovarian
cancer by 20 percent [19] .

Certain gynecological surgical procedures are also associated with a lower likelihood of ovarian
cancer. As examples, the risk of ovarian cancer is reduced by about 15 percent after tubal
ligation or hysterectomy with ovarian preservation. The protective mechanism of these
procedures may relate to impairment of ovarian function, causing anovulation, or protection from
exposure to exogenous carcinogens that enter the peritoneal cavity through the vagina. Tubal
ligation or removal of the uterus and cervix prevents possible upward migration of carcinogens
through the vagina, cervix, and fallopian tubes into the peritoneal cavity [19] .

Infertility may increase the risk of ovarian cancer [20,24,25] . A small increase in risk of ovarian
cancer has also been associated with endometriosis and particularly with endometriosis and
coexisting infertility [26,27] . Large meta-analysis and cohort studies have found that fertility
treatment does not independently increase cancer risk [24,28] .

A meta-analysis of over 20 case-control studies and a single large cohort study calculated an
estimated relative risk of ovarian cancer of 1.27 (95% CI 1.0-1.6) associated with prolonged use
of hormone replacement therapy (HRT) for 10 years or more [29] . Furthermore, an update of the
cohort study found a significant two-fold increase in the risk of ovarian cancer death associated
with use of HRT for more than 10 years [30] . Although it did not achieve statistical significance,
an increased risk with HRT was also seen in the Women's Health Initiative trial (hazard ratio 1.58,
95% CI 0.77-3.24) [31] . (See "Postmenopausal hormone therapy: Benefits and risks", section on
Ovarian and endometrial cancers).

RISKS AND BENEFITS OF SCREENING The potential benefit of screening is its ability to identify
ovarian cancer at a more localized and curable stage, leading to reduced mortality from the
disease. Three randomized controlled trials are in progress to determine whether earlier
diagnosis and decreased mortality (compared with no screening) can be achieved. Estimates of
the benefits of screening must be considered tentative at best until the results of these studies
are available.

The potential risks associated with screening for ovarian cancer must also be considered.
Although ovarian cancer is an important cause of cancer death, its incidence and prevalence in
the general population are relatively low. The problem of false-positive screening tests becomes
critically important in diseases with such a low prevalence. Since a positive screening test for
ovarian cancer would probably lead to surgery (either laparoscopy or laparotomy), a large
number of healthy women would be undergoing these procedures unnecessarily, unless the test
or sequence of tests was extremely accurate. These invasive procedures carry immediate
physical and psychological morbidity, as well as a small risk for serious complications; they also
account for substantial financial costs.

SCREENING TESTS Most experts feel that a screening protocol for ovarian cancer should have
a positive predictive value of at least 10 percent (that is, no more than nine healthy women with
false-positive screens would undergo unnecessary procedures for each case of ovarian cancer
detected). The predictive value of a screening test varies according to the prevalence of disease
in the population being screened and the specificity of the screening test. Manipulations of both
of these elements have been considered in efforts to design cost-effective screening programs. A
screening program that targets all women over age 50 would require a test with a specificity of
at least 99.6 percent (assuming a sensitivity of 80 percent) to achieve a positive predictive value
of 10 percent.

Ovarian tumors are occasionally detected on pelvic examination, although early stage tumors
are rarely found due to the deep anatomic location of the ovary. Thus, tumors detected by pelvic
examination are usually at an advanced stage and associated with a poor prognosis [32] . The
PAP smear may occasionally reveal malignant ovarian cells, but the test has a sensitivity for the
detection of ovarian cancer of only 10 to 30 percent [32] .

Tests that may be useful for screening in certain circumstances include measurement of the CA
125 tumor marker, ultrasonography, and combinations of the two.

CA 125 tumor marker Measurement of the serum concentration of the glycoprotein CA 125
antigen is the most widely studied biochemical method of screening for ovarian cancer. Serum
CA 125 values are elevated in over 80 percent of women with ovarian cancer. The average
reported sensitivities for early stage disease are 50 percent for stage I and 90 percent for stage II
[33] .

However, the specificity of CA 125 is limited. CA 125 is also increased in a variety of benign
conditions, such as endometriosis [34] , uterine leiomyoma, cirrhosis with or without ascites
[35,36] , and pelvic inflammatory disease, as well as in patients with other malignancies,
including endometrial, breast, lung, and pancreatic cancer [37] . The presence of pleural or
peritoneal fluid or disease involvement of a serosal surface, whatever the cause, also increases
CA 125 levels [38] . CA 125 levels are elevated in approximately 1 percent of healthy women
(show table 4) [39] , and fluctuate during the menstrual cycle. Mean CA 125 levels further vary
with ethnicity and smoking status (lower in non-White women and current smokers) and increase
with age [40] .

A prospective study of asymptomatic postmenopausal women found that an elevated CA 125

concentration was a powerful predictor of subsequent ovarian cancer risk [41] . In this study, the
relative risk of developing ovarian cancer within one year and five years was increased 35.9- and
14.3-fold, respectively, with a serum CA 125 concentration 30 U/mL.

Despite these impressive numbers, screening studies using single measurements of CA 125 have
not achieved favorable results. Studies of CA 125 in screening for ovarian cancer have focused
upon postmenopausal women, since menstrual cycle variations and the prevalence of benign
gynecologic conditions in premenopausal women would result in a substantially higher likelihood
of false-positive tests. Estimates of the sensitivity and specificity of the CA 125 assay are drawn
from three large screening studies in Sweden and England [42-44] and from numerous studies of
the test's performance in women with clinically diagnosed ovarian cancer [33,45] : The
sensitivity of the CA 125 assay (reference level 35 U/mL) for detecting preclinical disease ranges
from 70 to 80 percent. The specificity of a single CA 125 level in community screening studies of
postmenopausal women is 98.6 to 99.4 percent. Nevertheless, in an average-risk population with
a low prevalence of disease, the false-positive rate remains unacceptably high. In one study it
was estimated that there would be 30 false-positive tests for every ovarian cancer detected
[33] . The positive predictive value of annual CA 125 testing alone (3 percent) does not meet the
level required for screening postmenopausal women at average risk.

The definitive assessment of a cancer screening test and its impact on cancer mortality in a
randomized controlled trial is underway for CA 125 as a component of the Prostate, Lung,
Colorectal, and Ovarian Cancer Screening (PLCO) trial, scheduled for completion in 2013 [46] . In
the ovarian component of PLCO, 78,237 healthy women between 55 and 74 years of age were
randomly assigned to screening and control groups; 39,115 women were assigned to screening
with annual CA 125 and annual transvaginal ultrasound. Data from the baseline screen in 28,816
women found an abnormal CA 125 in 436 women (1.5 percent); the positive predictive value for
invasive cancer was 3.7 percent.

The change in CA 125 levels over time appears to be a more specific screening method. In one
study, the specificity reached 99.9 percent after redefining a positive test as a CA 125
concentration greater than 35 U/mL that doubles within six months [42] . Other reports have
found that the temporal pattern of CA 125 values or the use of tumor markers complementary to
CA 125 (eg, HE4) can increase the sensitivity with no loss of specificity [47] . A large prospective
study in 9233 postmenopausal women, with measurements of CA 125 at two or more times,
used a modeling method to calculate risk [48] . Compared with a specific cutoff value of CA 125,
the model improved sensitivity for detection of ovarian cancer from 62 to 86 percent when
specificity was fixed at 98 percent.

Thus, a single measurement of CA 125 is not an adequate screening tool for ovarian cancer. A
method based upon temporal changes in CA 125 levels could theoretically have sufficient
specificity for screening average-risk women, but this has not been proven. CA125 as a first line
test, followed by ultrasound as a second line test for a positive CA125 result, has shown to be
very specific and achieve a positive predictive value of 20% or greater [49] . The issue that
remains is whether CA125 is sufficiently sensitive to identify a large proportion of cases in early
stage disease. Studies are ongoing to identify additional markers to increase sensitivity while
maintaining the same high specificity.

Ovarian cancer symptom index An ovarian symptom index has been developed and proposed
for screening purposes [50] . The index is considered to be positive in women who report pelvic
or abdominal pain, bloating, increased abdominal size, difficulty eating or early satiety occurring
more than 12 times a month, with symptoms present for less than one year.

Among a group of 75 patients with pelvic masses subsequently diagnosed with ovarian cancer,
the symptom index had lower sensitivity for ovarian cancer than an abnormal CA 125 (64 versus
79 percent) although 11 percent had a positive symptom index and negative CA 125. The
combination of either a positive CA 125 or screening index identified 89 percent of the ovarian
cancers in this cohort. Among 254 high risk patients who served as controls, the symptoms index
was positive in 12 percent and CA 125 positive in 5 percent. The authors propose that a finding
of either a positive index or elevated CA 125 in high risk patients should lead to ultrasound
examination.

Lysophosphatidic acid The lipid lysophosphatidic acid (LPA) is associated with invasion of the
extracellular matrix in ovarian cancer. Serum LPA concentrations are elevated in 96 percent of
women with ovarian cancer, including 90 percent of those with stage I disease [51,52] . Studies
to evaluate the use of this and other biomarkers of disease are ongoing [53] .

Other tumor markers A variety of other tumor markers have been investigated for early
detection of ovarian cancer, including lysophosphatidic acid (LPA), CA 72-4, macrophage-colony
stimulating factor (M-CSF), osteopontin, inhibin, and kallikrein [54] . Studies are ongoing to
evaluate combinations of tumor markers complementary to CA 125 that could offer greater
sensitivity and specificity than CA125 alone. Examples of these studies include: OVX1, CA 125,
and macrophage-colony stimulating factor (M-CSF) were measured in 281 women with ovarian
tumors (175 malignant, 29 borderline, 77 benign) and 117 normal controls [55] . The panel was
significantly more sensitive than CA 125 alone for presence of any ovarian malignancy
(sensitivity 85 versus 80 percent) or stage I ovarian carcinoma (76 versus 66 percent). No single
marker performed better than CA 125. A composite index, incorporating four serum markers (CA
125, CA 72-4, CA 15-3, and M-CSF) has been developed using artificial neural network (ANN)
analysis as a modeling tool [56] . With a specificity set at 98 percent, the sensitivity of the index
for detecting invasive early stage ovarian cancer in a test set of 98 healthy women and 52
women with early cancer was 71 percent, compared to 43 percent for CA 125 alone. A fouranalyte assay for serum proteins (leptin, prolactin, osteopontin and insulin-like growth factor
[IGF]), when tested in serum from women with known epithelial ovarian cancer and in normal
controls, showed the following test attributes: sensitivity 95 percent, specificity 95 percent,
positive predictive value 95 percent, and negative predictive value 94 percent [57] . In
preliminary studies, this composite marker assay was also positive in women with breast and
uterine cancers, and its specificity for ovarian cancer is uncertain. Two additional markers (CA125 and macrophage inhibitory factor [MIF]) were added to the above four-analyte leptin,
prolactin, osteopontin and IGF, and tested as a six analyte mutiplex assay, comparing results for
women with newly diagnosed ovarian cancer and healthy controls [58] . The panel of six markers
had a sensitivity of 95.3 percent, specificity of 99.4 percent, positive predictive value of 99.3
percent and negative predictive value of 99.2 percent. A multicenter validation study is planned.
The test is commercially available in the United States, although its use for screening is not
recommended by the Society of Gynecologic Oncologists, pending results of validation studies
[59] . HE4 appears to have similar sensitivity to CA125 when comparing serum from ovarian
cancer cases to healthy controls, and a higher sensitivity when comparing ovarian cancer cases
to benign gynecologic disease [47] . Studies are underway to determine if HE4 significantly adds
to the sensitivity of CA125 while maintaining a high specificity.

Proteomics Proteomic patterns measured in serum may allow for the detection of early stage
ovarian cancer [60] . (See "Overview of gene expression profiling and proteomics in clinical
oncology", section on Proteomics). A study using mass spectroscopy proteomic analysis of serum
found that the procedure correctly classified all 50 samples from women with ovarian cancer
(sensitivity 100 percent) including 18 women with stage I cancers, and it correctly classified 63 of
66 samples from women without ovarian cancer (specificity 95 percent) [61] . However, this
study did not look at the performance of the proteomic analysis in a prospective screening
population, and the results have not been reproducible [62,63] .

Pelvic ultrasonography Transabdominal and, more recently, transvaginal ultrasonography have

been evaluated as potential first line screening methods for ovarian cancer. Transvaginal
ultrasonography (TVUS) allows better visualization of the ovaries (independent of body habitus)
and shorter examination times. There is disagreement about whether the addition of Doppler
imaging to ultrasonography improves diagnostic accuracy [64,65] .

The sensitivity of ultrasonography has ranged from 80 to 100 percent in studies of women with
clinically detected ovarian cancer and in several prospective screening studies [33] . The test's
specificity has ranged from 94 to 99 percent in screening studies, including two studies of
women with a family history of ovarian cancer [66,67] .

In screening studies in women at high risk of ovarian cancer, ultrasonography has performed
poorly in detecting early stage epithelial ovarian cancer. In the National Ovarian Cancer Early
Detection Program, 4526 women at high risk for ovarian cancer (based on a personal or family
history of ovarian or breast cancer, other cancer syndromes, or the presence of a BRCA
mutation) were screened with ultrasonography scans starting in 1990 [68] . After 12,709 scans
were performed, all ovarian, primary peritoneal, and fallopian tube cancers detected by
ultrasonography during the screening period were stage III; no early stage disease was identified.

Other studies of screening with ultrasonography in women at lower risk have shown somewhat
more favorable results. The largest prospective study of annual TVUS screening in 14,469
asymptomatic women (50 years of age and above, or 25 years of age and above with a family
history of ovarian cancer) is ongoing at the University of Kentucky [69] . An update from this
study reported that in 25,327 women screened with ultrasonography, screening detected 44
primary ovarian cancers; of these, 28 were stage I, 8 stage II, and 8 stage III, and 9 were serous
tumors of low malignant potential [70] . In a British study of ultrasound-based screening in 2,500
women with at least one family member with ovarian cancer, 12 ovarian cancers were identified;
11 of these were detected by ultrasonography and 64 percent were stage I [71] . It remains
possible that ultrasonography may be more effective in identifying early stage disease in lower
risk women than in women with hereditary cancer syndromes.

Multimodal screening Combination screening with serum CA 125 and ultrasonography, either
performed sequentially with patients receiving ultrasound only if the CA 125 is elevated, or
performing both tests in all patients, has been evaluated in multiple studies. A study of 22,000
postmenopausal women measured the serum CA 125 concentration in each; those with a
concentration 30 U/mL underwent ultrasonography [44] . Surgery was recommended for those
with an abnormal ultrasound. The screening protocol achieved a specificity of 99.9 percent and a
positive predictive value of 26.8 percent. Surgical exploration occurred in four women for every
case of ovarian cancer detected. A randomized controlled trial of annual multimodal screening in
the same group of 22,000 postmenopausal women has reported results of seven-year follow-up
[49] . Median survival in women with ovarian cancer in the screened group was 73 months,

compared with 42 months in the unscreened control group. While there was a trend toward
improved survival, no significant difference in mortality was detected. However, the study was
designed as a feasibility study and did not have sufficient power to detect mortality differences.
In a randomized controlled trial of 83,000 postmenopausal women in one prefecture in Japan,
42,000 women were invited to participate in annual screening with pelvic ultrasound and CA-125
[72] . The control group received usual care. There was no significant difference in the detection
of ovarian cancer, at an average follow-up of 9.2 years, between patients who received screening
(27 cases) and control patients (32 cases). There was a non-significant trend toward earlier stage
disease in the screened group. Thirty-three surgeries were performed to detect each case of
ovarian cancer. Mortality data are not yet available. Results have been reported from a
preliminary study from which a large prospective trial has been designed, and is currently
underway with an enrollment of 200,000 women 50 years of age in the United Kingdom (UK)
[73] . In the preliminary study, women were randomly assigned to screening by an algorithmic
strategy incorporating serial CA 125 measurements and secondary transvaginal ultrasound
(TVUS). Each woman's CA 125 profile was compared to the CA 125 profile of cases identified in
previous screening trials, to assess her risk of having undetected ovarian cancer. Results were
interpreted according to an individual's estimated risk profile; algorithmic guidelines determined
which women proceeded to TVUS. At one year follow-up from the baseline screen of 6532
women, three invasive epithelial ovarian cancers were detected (two stage I and one stage II).
The screening strategy had a positive predictive value of 19 percent and a specificity of 99.8
percent for invasive epithelial ovarian cancer. The Prostate, Lung, Colorectal, and Ovarian Cancer
Screening (PLCO) trial is another large trial underway; one component will evaluate mortality as
an outcome in postmenopausal women who have been randomly assigned to receive screening
with CA 125 and transvaginal ultrasound [46] . At baseline evaluation of 28,816 women, 1740
had either an abnormal CA 125 or ultrasound, and 34 had both. Nearly one in three women who
had a positive screening test underwent surgery (at the discretion of their physicians) but did not
have cancer. Among 570 women who had surgery, 29 tumors were found, of which 20 were
invasive. In this report, which reflects only prevalent cases found on initial screening, the positive
predictive value of having both tests abnormal was 23.5 percent. Twelve of the 20 invasive
tumors would have been missed if surgery were performed only if both tests were abnormal.

Preliminary results from follow-up testing after the baseline screens, that exclude baseline
prevalent cancers, provide a better assessment of the PLCO protocol [74] . Over the course of
three annual screening exams in 28,460 women, 32 invasive ovarian, fallopian tube, or
peritoneal cancers, and 4 tumors of low malignant potential, were detected. The positive
predictive value for invasive cancer among average risk women was 1.5 percent.

High risk women Reports from surveillance programs using multimodal screening in women at
high risk due to a genetic predisposition or family history have been disappointing: In one
surveillance program for women over 35 years of age with a family history of ovarian, breast,
colon, or endometrial cancer, or a personal history of breast cancer, 1261 participants were
screened with transvaginal sonography, color Doppler imaging, and CA 125 every one to two
years for a total of 6082 screens [75] . Although three stage I ovarian carcinomas were detected
by ultrasound, seven peritoneal serous papillary carcinomas with metastasis occurred despite
the screening intervention. In a smaller series of 383 women 18 years or older with a hereditary
breast and/or ovarian cancer syndrome who were screened annually for six years with

transvaginal ultrasound and CA 125, 20 had exploratory surgery for abnormal surveillance
results, detecting only one malignancy (metastatic breast cancer); two interval cancers of the
ovary and fallopian tube occurred despite surveillance [76] . Four years of screening with CA 125
and transvaginal ultrasound had a sensitivity of 40 percent and specificity of 99 percent in a
series of 312 women 35 years or older who were carriers for BRCA 1 or 2 (screened
semiannually) or members of a family with hereditary breast and/or ovarian cancer syndrome
(screened annually) [77] . Three out of the four early stage tumors found at prophylactic bilateral
salpingo-oophorectomy (n = 156) were in women who had normal CA 125 and ultrasound
findings. In a cohort of 888 women carriers of BRCA 1 or 2 mutations who underwent screening
with annual transvaginal ultrasound and CA 125, 5 of 10 incident cancers were interval cases
diagnosed in women who had had normal screening results three to ten months previously [78] .
Eight of the ten cancers were stage III at diagnosis. Preliminary data are available from the PLCO
trial through the first four screening examinations (baseline plus three annual screens with CA
125 and ultrasound) include results for high-risk women [74] . Women who underwent screening
(n=28,460) were stratified for risk by personal and family history: average risk (no history breast
or ovarian cancer), moderate risk (one first degree relative with breast cancer), or high risk
(family history of ovarian cancer, two relatives with breast cancer, or personal history of breast
cancer). After three post-baseline screening exams, the positive predictive value for abnormal
screening results was 2.8 percent in the highest risk group.

Poor results from intensive surveillance programs in women at high risk due to genetic
predisposition have led some experts to advocate other interventions to reduce risk in this
population, including use of the oral contraceptive pill and prophylactic oophorectomy.

Cost-effectiveness The cost-effectiveness of a multimodal screening protocol using annual CA

125 as a first-line test was less than $100,000 per year of life saved in one study based on
computer modeling [79] . More reliable data on cost-effectiveness awaits the completion of two
ongoing randomized trials in the United States and the United Kingdom.

RECOMMENDATIONS OF EXPERT GROUPS Because prospective screening trials are not yet
completed, and the impact of screening on ovarian cancer is still an open issue, no North
American expert groups recommend routine screening for ovarian cancer. A National Institutes of
Health Consensus Conference on Ovarian Cancer recommended taking a careful family history
and performing an annual pelvic examination in all women; other screening procedures such as
CA 125 testing and ultrasonography were recommended only for women with a presumed
hereditary cancer syndrome [80] .

The US Preventive Services Task Force [81] , the American College of Obstetricians and
Gynecologists [82] , the American College of Physicians [83] , and the Canadian Task Force on
the Periodic Health Examination [84] all recommend against routine screening for ovarian cancer
in asymptomatic women.

INFORMATION FOR PATIENTS Educational materials on this topic are available for patients.
(See "Patient information: Ovarian cancer screening" and see "Patient information: Genetic
testing for breast and ovarian cancer"). We encourage you to print or e-mail these topic reviews,
or to refer patients to our public web site, www.uptodate.com/patients, which includes these and
other topics.

RECOMMENDATIONS Clinical studies of ultrasonography and the CA 125 radioimmunoassay

have so far been too small to provide sufficient support to justify a policy of routine ovarian
cancer screening for most women. No definitive large randomized controlled trials have been
completed to show whether any screening strategy decreases mortality from ovarian cancer.
Nevertheless, a practical clinical approach to the issue of screening can be based upon
assessment of an individual woman's risk of ovarian cancer.

Women at average risk Screening for ovarian cancer with CA 125 or ultrasound is not currently
recommended for premenopausal and postmenopausal women without a family history of
ovarian cancer [85] . The predictive value of either test alone (less than 3 percent) yields an
unacceptably high rate of false-positive results and attendant morbidity and costs. Initial studies
suggest that CA 125 testing, serially or in combination with ultrasound, offers a promising
approach to screening asymptomatic postmenopausal women. However, in the absence of a
randomized trial demonstrating reduced mortality, no screening strategy can be recommended
at this time.

Women at higher risk In women with a family history of ovarian cancer, it is important to
differentiate those with a possible rare familial ovarian cancer syndrome and those with the more
common presentation of an isolated family member with ovarian cancer, without evidence of a
hereditary pattern. Risk assessment criteria for a high-risk familial pattern are shown in the table
(show table 2).

Lower risk family history Women with a family history but without evidence of a high-risk
pattern should be counseled about their individual risk (considering age, parity, and history of
oral contraceptive pill use), about the limited evidence for effectiveness of screening, and about
potential adverse effects. Postmenopausal women with this type of limited family history may be
reasonable candidates for screening because of an elevated risk of developing ovarian cancer.

For women who desire screening, a strategy of annual CA 125 testing with transvaginal
ultrasound in women with CA 125 levels above 30 U/mL is the approach that is best supported by
existing data.

High risk family history Women with a suspected hereditary ovarian cancer syndrome (show
table 2) should be referred to a genetic counselor for consideration of testing for BRCA1 and
BRCA2 mutations; consideration should also be given to referring Ashkenazi Jewish women with a
single family member with breast cancer before age 50 or with ovarian cancer. (See "Overview of
genetics in breast and ovarian cancer" and see "Genetic testing for breast and ovarian cancer"
and see "Overview of genetic assessment").

Women with a high-risk family history but who test negative for mutations can be managed
similarly to women with a lower risk family history (see "Lower risk family history" above).

Women who are found to have BRCA1 and/or BRCA2 mutations should be referred to a
gynecologic oncologist for follow-up. Protocols in clinical use for surveillance of such women
include combinations of pelvic examinations, CA 125 and other tumor marker measurements,
vaginal ultrasonography, and color Doppler imaging. The optimal interval for screening has not
been determined. Expert groups have recommended a six-month interval [86,87] , and this is a
reasonable option. However, the evidence indicate limited effectiveness of screening in this
population, and physicians and patients should not be falsely reassured by negative screening
test results.

Prophylactic oophorectomy at the completion of childbearing or by age 35 has generally been

recommended for women with hereditary ovarian cancer syndromes. Evidence from prospective
[88,89] and retrospective [87,90] studies of women with BRCA1 or BRCA2 mutations
demonstrated a substantial reduction in subsequent ovarian and breast cancers in women who
had prophylactic oophorectomy compared with those undergoing surveillance only. In addition, a
decision analysis found that a 30-year-old woman who carried a mutation in either the BRCA1 or
BRCA2 gene would on average gain 0.3 to 1.7 years of life from prophylactic oophorectomy [91] .
Gains in life expectancy declined with age at the time of surgery and were minimal for 60-yearold women, although there was little loss in life expectancy if the surgery was performed at age
40 rather than age 30. (See "Risk reducing salpingo-oophorectomy in women at high risk of
epithelial ovarian cancer").

When a hereditary ovarian cancer syndrome is suspected, the National Cancer Institute Cancer
Information Service (1-800-4-CANCER) can supply information about genetic services at cancer
centers supported by the Institute. In addition, the Society of Gynecologic Oncologists sponsors
the Women's Cancer Network at www.wcn.org. The site is designed to educate the public about
cancers of the reproductive system and includes a confidential questionnaire allowing users to
assess their individual risk of developing cancer of the ovary.

Risk factors for ovarian cancer

Familial ovarian cancer syndrome
Two or three relatives with ovarian cancer

Relative Lifetime
risk
probability,
percent*
Unknown 30 to 50
4.6
5.5 (15 if first
degree)
3.1
3.7 (5 if first degree)

One relative (first or second degree) with ovarian

cancer
No risk factors
1.0
1.8
Past oral contraceptive use
0.65
0.8
Past pregnancy
0.5
0.6
Infertility
2.8
Nulligravity
1.6
Past breast feeding
0.81
Tubal ligation
0.59
* Indicates probability for ovarian cancer in a 50-year-old woman.
Adapted from data in Carlson, K3, Skates, S3, Singer, DE, Ann Intern Med 1994; 121:124 and
Whittemore AS, R Harris, J Intyre, and the Collaborative Ovarian Cancer Group. Am J Epidemiol
1992; 136: 1184 and Gotlieb, WH, Baruch, GB, Friedman, E. Semin Surg Oncol 2000; 19:20 and
Ness, RB, Cramer, DW, Goodman, NT, etal. Infertility, fertility drugs, and ovarian cancer: a pooled
analysis of case-control studies. Am J Epidemiol 2002; 155:217.

Risk assessment criteria for inherited breast-ovarian cancer syndrome

Non-Jewish families
Any of the following:
One case of breast cancer <40 yo in a FDR or SDR*
One FDR or SDR with breast and ovarian cancer, at any age
Two or more cases of breast cancer in FDRs or SDRs if one is diagnosed at <S0 years
old, or is bilateral
One FDR or SDR with breast cancer at <50 years old, or bilateral and one FDR or SDR
with ovarian cancer
Three cases of breast and ovarian cancer (at least one case of ovarian cancer) in FDRs
and SDRs
Two cases of ovarian cancer in FDRs and SDRs
One case of male breast cancer in an FDR or SDR if another FDR or SDR has (male or
female) breast or ovarian cancer
Jewish families
Any of the following:
One or more cases of breast cancer <50 years old in an FDR or SDR
One or more cases of ovarian cancer at any age in a FDR or SDR
One or more FDRs or SDRs with breast cancer at any age, if another FDR or SDR has
breast and/or ovarian cancer at any age
One or more cases of male breast cancer in an FDR or SDR

* FDR: first-degree relative; SDR, second-degree relative. Adapted from Hampel, H, et al.
J Med Genet 2004; 41:81.

Estimated cancer risks associated with BRCA1 and

BRCA2 mutations
Type of cancer
Estimated lifetime risk in Estimated lifetime risk in
BRCA1 mutation carriers BRCA2 mutation carriers

Lifetime risk in
general population

Breast cancer (2)

47 to 66 percent*

40 to 57 percent*

12.5 percent

Contralateral breast
cancer (1)

Up to 65 percent

Up to 50 percent

0.5 to 1 percent per

year

Ovarian cancer (2)

35 to 46 percent

13 to 23 percent

1.5 percent

Colon cancer (1)

5 percent

Prostate cancer (1)

Not increased, or increased Not increased, or increased

very slightly
very slightly
Elevated (risk unknown)
35 to 40 percent

Male breast cancer (3)

0.2 to 2.8 percent

3.2 to 12 percent

0.1 percent

Pancreatic cancer (1)

<10 percent

<10 percent

1.3 percent

15 percent

* These ranges represent 95 percent confidence intervals derived from a meta-analysis of 10 individual studies representing both
high-risk and population-based cohorts. Many genetic specialists quote a broader range of lifetime risk for breast cancer BRCA1 (55 to
85 percent) and BRCA2 (50 to 85 percent) carriers.
1. Data compiled from Ford, D, et al. Am J Hum Genet 1998; 62:676, Struewing, JP, et al. N Engl J Med 1997; 336:1401, Antoniou, A, et
al. Am J Hum Genet 2003; 72:1117, Brose, MS, et al. J Natl Cancer Inst 2002; 94:1365, King, MC, et al. Science 2003; 302:643, Ford, D,
et al. Lancet 1994; 343:692, Breast Cancer Linkage Consortium. J Natl Cancer Inst 1999; 91:1310, Thompson , D, et al. J Natl Cancer
Inst 2002; 94:1358, Garber, JE, et al. J Natl Cancer Inst 2004; 96:2, and Liede,A, et al. J Clin Oncol 2004; 22:735.
2. Data from: Chen, S, et al. J Clin Oncol 2007; 25:1329.
3. Data from: Tai, YC, et al. J Natl Inst 2007; 99:1811