Possible hope for ovarian cancer screening

Since the discovery more than 30 years ago of a protein shed by tumor cells in the ovaries, researchers have tried unsuccessfully to use it for an ovarian-cancer screening test.

Now, a mammoth, long-awaited United Kingdom study has had some success by tracking rapid changes in blood levels of the protein, CA125, rather than simply elevations above a presumed normal.

A key to this screening strategy is an ovarian-cancer risk formula, or algorithm, developed over many years by a Harvard biostatistician with help from oncologists, including ones at Fox Chase Cancer Center in Philadelphia.

The British researchers caution that they won't have final data until late this year showing whether screening with the formula reduced deaths.

"We were able to safely and effectively deliver screening for over a decade" at 13 sites in England, Wales, and Ireland, study leader Usha Menon, a gynecological cancer researcher at University College London, said in a statement. "While this is a significant achievement, we need to wait . . . to know whether the cancers detected through screening were caught early enough to save lives."

Steven Skates, the biostatistician at Harvard and Massachusetts General Hospital, said he was hopeful that deaths were prevented. "But if not," he said, "then we'd have to conclude this approach doesn't help."

Even if the screening helps, it might not be embraced. The study, published Monday in the Journal of Clinical Oncology, involved a complicated, costly protocol for doing progressively more intensive diagnostic testing - added CA125 tests, ultrasounds, CT scans, MRIs, and surgery - on women who were flagged as being at risk by the algorithm.

"The cost of all the procedures to find tumors in a low-risk population would limit the public health value of this approach," speculated Mary Daly, a Fox Chase ovarian cancer researcher who has collaborated with Skates and hopes the approach will work.

Ovarian cancer has no distinctive symptoms, so only about 20 percent of cases are diagnosed early, when the disease is confined to the ovaries and readily curable with surgery. Although ovarian cancer is relatively rare - a one in 70 lifetime risk, compared with one in 8 for breast cancer - it remains one of the deadliest gynecological malignancies. Of the estimated 21,000 women diagnosed this year, 80 percent will die within five years.

More than 200 proteins that are unusually high or low in women with ovarian cancer have been studied as potential tumor "biomarkers." So far, none has been better than CA125.

"People have been predicting the demise of CA125 for the last 25 years," Skates said, "and yet it is still the primary ovarian cancer marker."

CA125 is currently used as a reliable marker of cancer recurrence in patients who have already been treated.

But as an early detection tool for the general population of postmenopausal women, it is problematic. About 15 percent of ovary tumors don't produce it, and many unrelated conditions - including bowel disease, endometriosis, and even pregnancy - can increase CA125 levels. That's why some women naturally have CA125 levels above the cutoff defined as normal (35 nanograms per deciliter of blood).

Researchers still don't even know the typical pattern of CA125 in cancer. Possibly, cancer has already spread beyond the ovaries by the time the protein becomes elevated.

"If you talk to most gynecologists, they say the [early stage] period is so short, we're never going to find it early using current methods," Daly said.

In 2011, a major clinical trial funded by the U.S. government found - as smaller studies had - that annual CA125 screening using a cutoff, even when combined with vaginal ultrasound imaging, did not catch the cancer earlier or save lives. Worse, since surgery is the only way to conclusively diagnose it, the risks of false alarms were high: 7 percent of women were sent for surgery that found no cancer, leaving 15 percent of them with surgical complications.

Expert groups currently recommend against CA125 screening, except for women with a genetic susceptibility to ovarian cancer.

Still, there was hope that using a statistical model to interpret CA125 changes over time would improve early detection, without so many unneeded surgeries. At least six studies that used Skate's ovarian-cancer risk algorithm - including one led by MD Anderson Cancer Center in Houston - were encouraging, but not large enough to dispel doubts.

The algorithm, which Skates has spent two decades refining, looks for unexpectedly large changes in a woman's baseline CA125 level, then compares her pattern to those of actual ovarian-cancer patients. When the algorithm identifies a woman at medium or severe risk of cancer, she is sent for more tests.

The just-published trial, begun in 2001, screened 46,236 women annually for a median of seven years. (Data from an unscreened comparison group are yet to come.)

In all, 640 screened women underwent surgery, which found 133 cancers - meaning about five women underwent surgery to find one with cancer. Thirty-five of the cancers (26 percent) were confined to the ovaries, or stage 1. And 20 (15 percent) were limited to the pelvis, or stage 2. Both rates are somewhat better than what is found now in the United States without any screening.

An additional 22 women were diagnosed within a year of their last screening because of symptoms such as bloating; this included a woman with aggressive, advanced cancer. None was detected by the algorithm.

Gynecological oncologist Karen Lu, who led the MD Anderson study, called the U.K. study "a huge tour de force."

"We will soon have an answer" about whether the screening strategy saves lives, she said. "Once we have it, we'll have to look at what is the benefit to an individual woman, and what's the benefit to the population as a whole."

215-854-2720 @repopter