Here’s a story about the way patients’ hopes and doctors’ good intentions can work together to stretch research into medical practice, for better or worse, even before the studies are finished.

Three and a half years ago, when she was 65, my sister learned that she had rectal cancer and would need radiation, extensive surgery and months of chemotherapy.

In the blur of frightening and numbing information that crashes down on an ordinary person who suddenly becomes a cancer patient, one thing seemed clear: Several doctors recommended Celebrex.

It is an anti-inflammatory drug originally developed as a painkiller, but studies had suggested that it could also shrink or prevent precancerous intestinal polyps, and some doctors reasoned that it could help patients who already had colorectal cancer.

In addition to the malignant tumor, my sister also had polyps. (Not every polyp turns into cancer, but virtually all intestinal cancers start out as polyps.)

Colorectal tumors are among the most common types of cancer in the United States, with about 105,000 new cases and 56,000 deaths in 2005, and more than a million survivors.

My sister’s oncologist, a private practitioner, prescribed high doses of Celebrex, 800 milligrams a day. Her surgeon, a department head at a major cancer center, seemed unenthusiastic about the drug, but O.K.’d it. So my sister took it.

Meanwhile, I asked two other experts, researchers at cancer centers, about Celebrex. They disapproved. The drug was being studied in people with polyps, not cancer patients. There was no proof it would help people like my sister, they said. Doctors should wait for data instead of jumping the gun. She shouldn’t be taking the drug. One was so adamant that he offered to talk to her himself.

But my sister liked and trusted her oncologist. I decided not to meddle.

The combined treatments worked. She made an amazing recovery, and the cancer has not come back. At the oncologist’s urging, she kept taking Celebrex even after she finished chemotherapy.

But then, in December 2004, after she had been taking the drug for 21 months, she got a call from her oncologist, telling her to stop immediately. A study had found that Celebrex could increase the risk of heart attacks. Vioxx, a similar drug, was taken off the market because of a stronger link to heart problems. My sister quit taking Celebrex.

But what about the cancer? She was doing so well. Had the Celebrex been helping her? Did the heart risk really outweigh the possible benefit, even in cancer patients or people at high risk?

The answer is still not entirely clear.

Last week, two studies and an editorial about Celebrex and intestinal polyps were published in The New England Journal of Medicine.

The studies were paid for in part by Pfizer, which makes Celebrex, and company scientists were among the authors of the studies.

Each study had about 2,000 participants. People who had had polyps were given either the drug or a placebo and then followed for three years. Both studies found that those taking Celebrex had significantly fewer polyps. The drug seemed especially good at preventing the type of polyps most likely to turn into cancer. But there were increased heart problems in the Celebrex group.

In one study, among patients taking 400 milligrams once a day, 33.6 percent formed polyps, compared with 49.3 percent in the placebo group. But 2.5 percent taking the drug had heart problems, compared with 1.9 percent taking placebos.

In the second study, 60.7 percent in the placebo group formed polyps. But the rate was only 43.2 percent in those taking 200 milligrams of Celebrex twice a day, and 37.5 percent in those taking 400 milligrams twice a day.

The heart risk from Celebrex also increased in the second study: it was 2.6 percent in those who took 200 milligrams twice a day and 3.4 percent in those who took 400 twice a day, compared with only 1 percent in the placebo group.

In both studies, some of the heart problems were fatal.

“For the average patient who’s had colon polyps at colonoscopy, I think the cardiac risks outweigh the benefits of this drug,” said Dr. John R. Saltzman, an author of one of the studies and director of endoscopy at Brigham and Women’s Hospital in Boston. But he added that Celebrex might still be worthwhile for a small subset of patients, those prone to fast-growing, high-risk polyps — provided they had no risk factors for heart disease.

“We’re dealing with two diseases that will kill you,” Dr. Saltzman said. “You have to weigh the risks and benefits very carefully when you’re dealing with such serious diseases.”

The editorial said Celebrex had “no role” in cancer prevention in the general population or in people who had polyps that were not part of a hereditary cancer syndrome. Its authors also said that people at risk for colon cancer because of a tendency to form polyps were better off having regular colonoscopies to remove the polyps and taking low-dose aspirin, which can lower the risk of heart attacks and also give a little protection against polyps, though not as much as Celebrex.

What about people like my sister, who have already had cancer? There is no data on them. Cancer patients were not part of the research, and the studies did not go on long enough to study the incidence of cancer.

Despite the lack of evidence, Dr. Saltzman said he was not surprised to hear that oncologists had been prescribing Celebrex for cancer patients anyway.

“What you’re dealing with is people who want to do the right thing,” he said. “They’re extrapolating data from one situation to another. It’s done all the time in practice. What we practice is part science and part art. Most physicians want to do the right thing for their patient.”

He added that when a drug looks promising, many doctors will recommend it even before the research is finished, especially if they’re frustrated because they have nothing else to offer.

Dr. Bernard Levin, a director of the second study and vice president for cancer prevention and population sciences at the M. D. Anderson Cancer Center in Houston, said that despite the drawbacks of Celebrex, the findings were still encouraging because they showed that a drug could reduce polyp formation.

The task now, he said, is to find one that can do it safely, and maybe even lower the risk of heart disease at the same time. He also said that his study found such a small heart risk from a 400-milligram dose twice a day that it deserved further study.

“This may be relevant for tomorrow’s patient,” Dr. Levin said. But for today’s patients, he said, “we’re sort of left hanging.”