CHICAGO — The experimental drug PLX4032, which reverses the effects of a mutation found in certain tumors, is considered a prime example of the “targeted” cancer therapies of the future. The drug — the subject of a three-part series in The New York Times in February — produced seemingly miraculous results in some patients with melanoma.

But when the same drug was tried in patients whose colorectal tumors had the same mutation, there was barely any effect, researchers reported here last week at the annual meeting of the American Society of Clinical Oncology.

Same drug, same mutation, different result.

The findings serve as another reminder of how devilishly complex cancer can be and how much more remains to be understood. They provide a sobering check on the enthusiasm that had been building about the targeted therapies, which act to block particular abnormalities that spur tumor growth.

“This is one of the few warning shots over the bow,” said Dr. Scott Kopetz of the M.D. Anderson Cancer Center in Houston, who led the colorectal cancer study.

Indeed, even as researchers at the meeting hailed the latest drug successes, they acknowledged that the path forward was not as easy as once envisioned. Many targeted therapies are failing in clinical trials.

“We’ve gone through a very rapid period of high expectations, maturation and disappointments,” said Dr. J. Leonard Lichtenfeld, the deputy chief medical officer of the American Cancer Society. “I think there was almost a naïveté that if we could find the target, we would have the cure.”

Dr. Kopetz said that after a period of huge gains that culminated with the approval in 2004 of two targeted drugs, Avastin and Erbitux, progress against colon cancer had stalled. “In the past five years I don’t think we’ve made any headway,” he said.

One of the stars of the oncology conference was Pfizer, whose experimental drug crizotinib shrank tumors in nearly all lung cancer patients whose tumors had a particular genetic abnormality. The abnormality was discovered only three years ago, and the drug is moving toward the market.

“It is based on biology and just speaks to the power of understanding the cancer cell,” Dr. Mark G. Kris, a lung cancer specialist at Memorial Sloan-Kettering Cancer Center in New York, said at a news conference.

But Pfizer has had numerous, if less publicized, failures in trials for other targeted therapies. Its kidney cancer drug, Sutent, did not work as a treatment for breast, colon and liver cancers. Another drug that blocks a protein called the insulin-like growth factor receptor — a hot target among drug companies — failed as a lung cancer treatment.

“I was really surprised when the negative results came out,” said Dr. Mace L. Rothenberg, an oncologist who leads Pfizer’s cancer clinical trials.

Conventional chemotherapy typically works by killing rapidly dividing cells. That can mean the tumor but also certain normal cells, causing side effects like hair loss, nausea, diarrhea and anemia.

Targeted therapies, by contrast, aim at proteins that are somehow contributing to tumor growth or survival but, ideally, are not found in healthy cells. That could translate into more effectiveness with fewer side effects.

Numerous targeted drugs are in use and are improving treatment. The paragon is Gleevec, which can keep chronic myeloid leukemia in check for years. Others include Tarceva for lung cancer and Nexavar for kidney and liver cancers.

But the drugs have their own side effects. And in many cases, they work best when used with chemotherapy, so patients cannot escape the side effects of chemo.

Enthusiasts for the targeted drug have been saying for years that tumors will eventually be characterized by their molecular profiles — which mutated genes they have — rather than where in the body they occur. Names like breast cancer and lung cancer will be supplanted by terms like B-RAF-positive or EGFR-positive tumors. And drugs will be chosen based on that profile, the way antibiotics are generally selected based on the pathogen that is causing the infection, not on where in the body the infection occurs.

Massachusetts General Hospital, for instance, is running a clinical trial testing a drug from AstraZeneca on any type of cancer — providing it has a mutation in the gene B-RAF, the same gene that is the target of PLX4032.

But the test of PLX4032 in colon cancer suggests that the location of the tumor still does matter, that it will not be just a case of looking at the target. There are other examples as well. Erbitux and Vectibix do not work in colon cancer patients with a mutation in a gene called K-RAS. But the relationship between the mutation and the effectiveness of Erbitux does not seem to hold in lung cancer.

Experts say that in certain cancers a genetic mutation might be present but is not really driving the cancer, as it might be in other types of cancer.

“You’ve got to target a specific disease, not a biomarker,” said Dr. Mark J. Ratain, an oncologist at the University of Chicago.

Even within the same type of cancer, there can be differences. Researchers reported at the conference that Erbitux, also known as cetuximab, did not prolong survival after surgery for early stage colon cancer, even though the drug works for metastatic colorectal cancer. The failure occurred even though all the patients in the trial were screened to make sure they did not have the mutated K-RAS gene that would have rendered the drug ineffective.

Another study found that in more than 12 percent of cases, tumors that spread to the liver from the breast differed from the original tumor in markers widely used to help determine treatment — estrogen receptor, progesterone receptor or Her2.

The physical characteristics of the drug also matter. Avastin works on the outside of cells to block the action of a protein called VEGF that spurs the formation of blood vessels that nourish tumors. It has prolonged survival in colon cancer and lung cancer.

But so-called small-molecule drugs that try to block the action of VEGF by working on the insides of cells have failed in many trials for lung cancer and colon cancer.

Still, researchers say the setbacks should not be surprising or discouraging but rather part of an effort to learn what makes cancer tick.

“I don’t find it disappointing, said Dr. Robert J. Mayer, a gastrointestinal cancer specialist at the Dana-Farber Cancer Institute in Boston. “I find it intriguing.”

Many researchers say the path forward is to go from merely targeted to truly personalized therapy.

Some leading genetic scientists from Harvard, M.I.T. and the Broad Institute recently helped start a company, Foundation Medicine, that plans to develop a comprehensive test profiling a patient’s tumor, to help doctors personalize therapy.

Some major cancer centers are already doing this on their own, particularly for lung cancer.

At Sloan-Kettering, tumor samples from patients with a certain type of lung cancer are tested for 91 mutations in eight genes. In half the cases, the tests have found a mutation that drives the cancer. In 18 percent of cases, the findings influence the choice of therapy, Dr. Kris said.

The drawback of such a personalized approach is that it will take years to develop drugs for each small subset of patients.

“I think the era of one drug for a whole disease type has passed,” said Dr. Rothenberg, of Pfizer. “We’re going to go where the science leads us. We just have to get smarter about it.”