American Association for Clinical Chemistry
Better health through laboratory medicine
July 2010 Clinical Laboratory News: The Diagnostic Dilemma Of Ovarian Cancer

CLN Banner Logo

July 2010: Volume 36, Number 7

The Diagnostic Dilemma Of Ovarian Cancer
What’s the Answer?

By Genna Rollins

If ever a disease was crying out for a screening biomarker, it would be ovarian cancer. This ‘silent killer’ is the most lethal gynecologic malignancy and the fifth leading cause of cancer death in women. While women diagnosed in early stages have a better than 90% 5-year survival rate, unfortunately, the majority first learn they have the disease when it is advanced, and only have about a 30% survival rate. Despite rigorous efforts by an army of dedicated researchers over decades, no single biomarker, panel, or combination serum-imaging strategy has cleared the sensitivity and specificity performance hurdles for this relatively rare disease. Given its estimated lifetime risk of 1 in 70, any screening strategy would need to have a minimum specificity of 99.6% and sensitivity of at least 75% to avoid an unacceptable level of false-positives and achieve a 10% positive predictive value (PPV). Clinicians and researchers are equally frustrated with the lack of tangible progress in meeting these stringent diagnostic criteria.

“Three-quarters of women today present as they did in 1950—with advanced stage cancer,” said David Fishman, MD, professor of obstetrics, gynecology and reproductive sciences and director of gynecologic oncology research at Mount Sinai School of Medicine in New York City. “Right now, early stage ovarian cancer is found by luck, not skill. If a woman has a tumor the size of a basketball that hasn’t spread, that’s not because the doctor’s clinical acumen is so good. It’s because that cancer was not one that was going to spread.”

Ovarian Cancer Stage Distribution and 5-year Relative Survival

Ovarian cancer 5-year survival rates are quite high when the cancer is detected in early stages, but most women learn they have the disease when it is advanced, decreasing their 5-year survival odds.

Stage at Diagnosis*
Stage Distribution (%)
5-year Relative Survival (%)
Localized (confined to primary site)
Regional (spread to regional lymphnodes)
Distant (cancer has metastasized)
Unknown (unstaged)

*1999–2006, All Races

Source: National Cancer Institute Surveillance Epidemiology and End Results

Incremental Gains

Despite the seemingly negative outlook, many investigators—Fishman included—remain optimistic that breakthroughs are on their way, albeit down the road a bit. One who shares his view is Daniel W. Chan, PhD, DABCC, FACB, professor of pathology, oncology, radiology and urology, and director of the clinical chemistry division and Center for Biomarker Discovery at Johns Hopkins Medical Institutions in Baltimore. “Because of the difficulty and complexity associated with this disease, there’s plenty of opportunity, but it’s going to take more time. Many of us were hoping to hit a home run, but things don’t happen that easy and fast. Any sort of incremental improvement will be beneficial to patients and clinicians,” he observed. Chan’s lab performed the initial research that lead to development of Vermillion’s OVA1 assay, which the FDA cleared in September, 2009, the first proteomic in vitro diagnostic multivariate index assay to be so designated. The test uses five biomarkers—cancer antigen 125 (CA 125), transthyretin, apolipoprotein A-1, beta2-microglobulin and transferrin—and proprietary software to determine prior to surgery the likelihood that a woman’s ovarian mass is malignant.

OVA1 is an example of the incrementalism Chan sees as the ultimate key to unlocking ovarian cancer diagnostics. “The OVA1 panel was chosen to improve the performance of CA 125, which it does mainly by increasing the sensitivity of the test. But it does not improve specificity, and in fact, specificity suffers,” he indicated. Chan also pointed out that OVA1 has not been sanctioned for use as a screening test.

The Old Standby: CA 125

CA 125, a glycoprotein found on the surface of epithelial cells, is the most studied serum biomarker in ovarian cancer. Various professional guidelines—including NACB’s Laboratory Medicine Practice Guidelines (LMPG) on the Use of Tumor Markers in Testicular, Prostate, Colorectal, Breast, and Ovarian Cancers—recommend using CA 125 for prognosis, detection of recurrence, monitoring therapy, differential diagnosis in suspicious pelvic masses, and as part of a strategy for early detection in hereditary syndromes. However, none have endorsed the assay as a screening test.

“CA 125 has been kicking around for decades and there’ve been a lot of studies and various approaches of looking at its performance. The general conclusion is that although it does detect some tumors before they’re clinically apparent, there’s not convincing evidence that it’s usefully detecting them early. And most of the ones detected early are the non-lethal kind,” observed Patrick Brown, MD, PhD, professor of biochemistry at Stanford University in Palo Alto, Calif.

CA 125 also has the disadvantage of not being specific to ovarian cancer. It is elevated in numerous other cancers such as colon, uterine, fallopian, and pancreatic, as well as many non-cancerous conditions, including pregnancy, liver disease, benign ovarian cysts, and endometriosis. Despite these shortcomings and lack of recommendations by professional groups, many clinicians currently use CA 125 levels in assessing women suspected of having ovarian cancer, according to experts. “It’s a tool that seems to have some clinical value but it’s being used in ways it was never approved for. Unfortunately, it’s being applied in situations where the test isn’t that accurate,” said Fishman. “In my opinion, it’s not a major player in the biology of the disease. I think it’s an artifact of an inflammatory process, and as we become more sophisticated and understand more of the mechanisms of early stage cancer, we’ll be able to detect proteins or earlier inflammatory markers.”

The NACB LMPG also identified several analytical issues with CA 125 assays, including antibody, calibration, and reagent differences. “Second-generation CA 125 assays use an additional antibody—M11— rather than just OC 125,” noted Chan. “It’s important to find out which reagent kits use which antibodies, because a company just saying it has a CA 125 assay does not mean they use the original antibody.” The NACB LMPG called for an international standard for CA 125 and cautioned that until one is developed, values between different methods are not interchangeable and patients who are serially monitored should be re-baselined if there is a change in methodology.

Did Screening Trials Provide Answers?

Despite the knocks against CA 125, it has figured prominently in the only major screening trials of ovarian cancer, all of which used levels of this biomarker, along with transvaginal ultrasound, in different testing strategies. Two of those trials—the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) and the U.S.-based Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening trial—are ongoing, but published interim results in 2009 (Lancet Oncol 2009;10:327–40; Obstet Gynecol 2009;113:775–82). The third, the Shizuoka Cohort Study of Ovarian Cancer Screening (SCSOCS), concluded follow-up in 2002 and reported results in 2008 (Int J Gynecol Cancer 2008;18:414–20).

Results from two of the trials were not exactly ringing endorsements of CA 125. SCSOCS, which assigned participants in the intervention arm to receive annual ultrasound and CA 125 testing, found no significant difference in the number of ovarian cancers detected among the screening and control groups. Although more stage I cancers were found in the screening compared with the control arm, the difference was not statistically significant. PLCO also involved annual transvaginal ultrasound and CA 125 tests, following by two additional rounds of CA 125 screening. The researchers reported in their interim results a PPV of only 1–1.3% across four screening rounds. Of greatest concern to the investigators, nearly three-quarters of screen-detected cancers were late stage and the ratio of surgeries to screen-detected cancers was nearly 20:1. “The only sure way to rule out ovarian cancer in the event of a positive test is a surgical procedure—there’s no biopsying of the ovary like there is with breast tissue following a positive mammogram. So the implications of a positive test are pretty substantial,” said lead author Edward Partridge, MD, professor and director of the Division of Gynecologic Oncolo-gy and director of the Comprehensive Cancer Center at the University of Alabama at Birmingham.

Interim results from UKCTOCS appear to shed a different light on the utility of CA 125. This three-arm trial assigned participants to usual care, annual CA 125 testing and transvaginal ultrasound scanning if CA 125 levels were abnormal, or annual transvaginal ultrasound scanning alone. In contrast to both SCSOCS and PLCO, UKCTOCS reported sensitivity of 89.4%, specificity of 99.8%, and PPV of 43.3% for the multimodal strategy combining annual CA 125 testing with sequential ultrasound. Significantly, 48% of invasive cancers were identified in stage I or II.

That UKCTOCS found a much higher percentage of early stage cancers may be due to its use of an algorithm that incorporates minor changes in CA 125 overtime as opposed to reporting values above or below a cutoff of 35 U/mL, according to the study’s lead author, Usha Menon MD, professor of gynaecological oncology and head of the Gynaecological Cancer Research Centre at University College London. “We feel the absolute value is important, but the trend is even more important,” she explained. “You could have a flat value of 100 and still not have ovarian cancer, or you could have a value rising from 8 to 15 that would make us really worried, even though it’s all in the normal range.” Researchers at the University of Texas MD Anderson Cancer Center in Houston also recently reported encouraging preliminary findings from a smaller screening trial that evaluated rising CA 125 levels in concert with transvaginal ultrasound.

Menon and her colleagues developed and refined their algorithm, which incorporates baseline and trend CA 125 levels and age, based on data from earlier studies they conducted. This approach is so unusual that the research team had to have a calibrant created for the assay they used in the study. “The company we work with was not interested in precision at the 8-to-12 range, for example, because calibrants normally are 30, 40, or higher. Ours is set at around 9 or 10 and we keep the CV very tight around that because we are looking at very subtle rises in CA 125,” she explained.

Despite these positive interim results, Menon was quick to caution that UKCTOCS is ongoing and that mortality data, which will be critical to establishing any screening strategy, won’t be available until 2014. Partridge agreed that a mortality benefit is the crucial hurdle for any screening modality. “We can’t stress enough that a difference in mortality has to be the end point. Anything less than that is a problem,” he said. Based on PLCO outcomes thus far, Partridge has a guarded outlook about the prospects for a screening strategy involving CA 125. “The bottom line for me as I look at our interim results is, it’s difficult to believe it’s going to impact mortality. Is there a biomarker out there that’s as yet undetected that’s going to make a difference? One would hope so because ovarian cancer is such a deadly disease,” he said. PLCO mortality data should be available within the next 6 months, according to Partridge.

Promising Leads

Given the challenges associated with CA 125, researchers are actively pursuing other biomarkers and avenues of investigation. The NACB LMPG identified 20 other currently available serum markers for ovarian cancer, but only CA 125 has been adopted in clinical use (See Table, below). Earlier this year, Abbott achieved European CE marks and in June received FDA clearance for its human epididymis protein 4 (HE4) assay to monitor recurrence or progression of epithelial ovarian cancer, the most common form of the disease.

Serum Markers for Ovarian Cancer

Numerous candidate serum markers are being investigated for use in detecting, differentially diagnosing, predicting prognosis, and monitoring ovarian cancer. The most promising are listed below.

Cancer Marker
Proposed Uses
Phase of Development
CA125 Differential diagnosis of pelvic masses; tumor monitoring Accepted clinical use
Her-2/neu Tissue marker for prognosis prediction and treatment outcome prediction Evaluation
Akt-2 Tissue marker for prognosis prediction Research/Discovery
Inhibin Detection Evaluation
HLA-G Differential diagnosis Research/Discovery
TATI Tumor monitoring Research/Discovery
CASA Tumor monitoring, prognosis prediction Research/Discovery
TPA Tumor monitoring Research/Discovery
CEA Tumor monitoring Research/Discovery
LPA Detection Evaluation
PAI-1 Prognosis prediction Research/Discovery
IL-6 Prognosis prediction Research/Discovery
Kallikreins 5,6, 7, 8, 9,10, 11,13,14,15 Differential diagnosis, tumor monitoring, prognosis prediction Research/Discovery
hCGβcf Prognosis prediction Evaluation
Prostasin Differential diagnosis Research/Discovery
Osteopontin Tumor monitoring Research/Discovery
HE4 Differential diagnosis of pelvic masses; monitoring therapy In clinical use in some centers
MAPK Tissue marker for prognosis prediction Research/Discovery
IGFBP-2 Prognosis prediction Research/Discovery
RSF-1 Prognosis prediction Research/Discovery
NAC-1 Prognosis prediction Research/Discovery

Abbreviations: TATI, tumor-associated trypsin inhibitor; CASA, cancer-associated serum antigen; TPA, tissue polypeptide antigen; CEA, carcinoembryonic antigen; LPA, lysophosphatidic acid; PAI-1, plasminogen activator inhibitor-1; IL-6, interleukin-6; hCGβcf, urinary β-core human chorionic gonadotropin; HE4,human epididymis protein-4 MAPK, mitogen-activated protein kinase; IGFBP-2; insulin-like growth factor binding protein–2.

Source: NACB LMPG on the Use of Tumor Markers in Testicular, Prostate, Colorectal, Breast, and Ovarian Cancers, Clin Chem 2008; 54:e11–79

Among the various lines of investigation being pursued, Chan believes glycomics is one of the most promising. “We know there is heterogeneity of the glycosylated forms of proteins in the cancer process. PSA in prostate cancer is a good example. I believe the same kind of strategy might work in ovarian cancer. We might find glycosylated proteins that would give us the sensitivity and specificity we need,” he indicated.

Fishman has proposed a model that would incorporate proteomics and contrast-enhanced ultrasound (Am J Roentgenol 2010;194:349–54). “As proteomics matures, we’ll be able to detect some of the proteins that are involved very early in ovarian cancer development. Our ultimate goal is to develop high throughput, low-cost assays that would evaluate a panel of biomarkers,” he explained. “The contrast-enhanced ultrasound, we believe, helps delineate the microvascular changes associated with early tumor development. If this technique is validated, it would be an inexpensive technique that can complement the evaluation of any woman who had an abnormal blood test.”

Recently Menon and colleagues identified a panel of four markers from 96 candidate proteins using Luminex’s multiplex xMAP bead-based immunoassays that the authors suggest could be used as a first-line test in a two-step strategy for early detection (J Clin Oncol DOI: 10.1200/JCO.2008.19.2484). CA 125, HE4, carcinoembryonic antigen, and vascular cell adhesion molecule-1 provided the highest diagnostic power of 86% sensitivity for early-stage and 93% for late-stage ovarian cancer at 98% specificity.

Many observers believe a panel of serum markers will be required because of the heterogeneity of ovarian cancer and the signal-to-noise problem associated with detecting tiny tumors. “It’s easy to find proteins that are substantially over-expressed in ovarian cancer than virtually any other tissue. The problem is, the tumor size at the time you need to detect it is on the order of one-millionth the mass of your body,” explained Brown. “All it takes is a relatively low level of expression in some other tissue and that protein will greatly exceed the amount produced by the tumor.”

Brown conducted a complex modeling analysis for the growth, progression, and detection of occult ovarian cancer and concluded that the tumors researchers would need to detect to achieve 50% sensitivity are more than 200 times smaller than the clinically apparent serous cancers typically used to evaluate performance of candidate biomarkers (PLoS Med 6(7): e1000114. doi:10.1371/journal.pmed.1000114). “That’s fundamentally doable, but it’s beyond the capabilities of any test that’s currently available, proposed or described,” said Brown. “On the other hand, it gives us a defined target for developing a test. It tells us what the challenge is.” The good news for researchers is that both PLCO and UKCTOCS have sizable specimen repositories available for evaluating candidate biomarkers.

What’s the Tumor Lead Time?

Brown believes many ovarian cancer-related biomarker studies have missed the mark by not targeting the serous form of the disease, which accounts for 60% of epithelial ovarian cancer and more than 80% of ovarian cancer deaths. His analysis revealed that these tumors spend approximately 4 years as in situ stage I or II cancers and about 1 year as stage III or IV before becoming clinically apparent. Other researchers have noted a dichotomy in the natural history of ovarian cancer. For instance, SCSOCS lead investigator Hiroshi Kobayashi, MD, PhD, described two distinct intervals between normal and elevated CA 125 levels. In nonserous-type cancers, the mean interval was 3.8 years, but in serous-type, it was just 1.4 years (Int J Clin Oncol 2009;14:378–382).

Kobayashi is among the investigators who have suggested that serous ovarian cancer may develop de novo. Indeed, a model for ovarian cancer pathogenesis has been proposed in which Type I slow-growing tumors develop from precursor borderline lesions and Type II aggressive tumors arise de novo. Should this model be validated, it could provide a pathway for future biomarker development, according to Partridge. “I’m not convinced the stage ones we’re detecting in these trials are really substantial contributors to mortality,” he said. “It may take a combination of two things—a biomarker that detects the disease and a second test that tells you how aggressive it is. You’d put those two pieces of information together to know whether there was a cancer and whether or not it was life-threatening. Then you might have something.”

Other researchers are not so sure about this model, and in fact, there are competing theories about whether the disease originates in the fallopian tubes or in the ovarian epithelium. “We don’t even agree that ovarian cancer starts in the ovary, much less that there’s a detectible precursor lesion. Until you have that kind of information, it’s hard to imagine developing a suitable test or exam,” contended Jeff Boyd, PhD, chief scientific officer at Fox Chase Cancer Center in Philadelphia.

Boyd and his colleagues recently conducted a detailed molecular genetic and morphologic analysis of normal and cancerous ovarian tissues that seemed to implicate ovarian inclusion cysts as the locus of precursor lesions (PLoS One 5(4): e10358. doi:10.1371/journal.pone.0010358). The researchers found aneuploidy and a higher cell proliferation index in histologically normal cells adjacent to dysplastic cells. “We were able to show genetic alterations in the same genes existed in normal, dysplastic and cancer cells adjacent to each other. This implies a clonal evolution because the chances of having the same mutation in the same genes in all three cell regions by chance is virtually nil,” he explained.

As the hunt continues for that special combination of biomarkers that will at last hold the key to early ovarian cancer diagnosis, Fishman sounded a cautionary note. “The fear all of us have is that there’ll be a panel of biomarkers that suggest you have cancer but we have no clue, no imaging technology, to pick it up because it’s too early,” he said. “Everybody is in such a hurry to create something, but let’s get it validated at the level the FDA would want. That way everyone can feel that even if we make a mistake, we’re doing it in the patient’s best interest.”

Dr. Chan serves on the scientific advisory board of Vermillion, and his employer licensed to Vermillion the technology involved in the OVA1 assay.