Ever since the day Marvin Williams died--

Nov. 23, 1979 -- his young widow wondered why the hospital had not sent her the autopsy report she had requested.

Susan Williams had kept a journal of her husband's two-year fight with leukemia, and she prided herself in knowing each detail of the treatment. But there was still one question the doctors at Houston's M.D. Anderson Hospital had not answered -- why her husband's final weeks had been so violent and ugly. She wondered why his 27-year-old body had been convulsed with constant seizures, why he struggled for each breath, and why his mind, once so keen, had failed to recognize even her.

What Susan Williams was not told was that the doctors treating her husband had given him an "accidental overdose" of an experimental drug.

The experience of Marvin Williams is not simply the story of an accident. It is also a story of deception at one of the many prestigious cancer centers in this country where thousands of patients have become the subjects of drug experiments sponsored by the National Cancer Institute.

Few of these patients have been subjected to an overdose. But, as was the case with Williams, most know little about the drugs they are given and understand even less about the risks of serious side effects. The Williams story and other case studies presented today show that patients are often unaware of the world of research they have entered in desperation: a world where some drug experiments go on for months and even years after the drug has failed to show it fights cancer; a world where some patients who might otherwise be at home and walking are instead bedridden with side effects; and, as one case study will show, a world where rivalry between researchers sometimes obscures the best treatment.

Hospital records show that Williams began taking an experimental drug called Deoxycoformycin for his advanced leukemia on Oct. 31, 1979, after talking it over with his wife and deciding to give it a try. But after two series of treatments, Williams was taken off the drug because his leukemia was not responding.

What happened next is recalled by Dr. Robert Benjamin, section chief at the Houston hospital: On Nov. 13, 1979, Dr. Michael Keating, Williams' chief physician, instructed one of his "fellows" to give Williams a standard anticancer drug, Daunorubicin.

The order was misread.

According to Benjamin, Williams was given a continuous infusion of the experimental drug Deoxycoformycin again, this time at twice the dose.

"Daunorubicin and Deoxycoformycin, when written out in doctors' language, look awfully similar," said Benjamin, who had referred to the incident as an "accidental overdose" in an April 23, 1981, letter to NCI. "None of us would have picked that dose for Deoxycoformycin," Benjamin said in an interview.

The error was discovered 12 hours after infusion began. It was the morning of Nov. 14, 1979. Dr. Keating was "on rounds," Benjamin recalled. As Keating approached Williams' bed he noticed the solution dripping into Williams' veins was clear, not red, the color of Daunorubicin.

Instead of halting the drug, Benjamin and Keating said they decided to continue it because Williams' leukemia was showing a "dramatic response."

"He Williams was totally aware after he started the Deoxycoformycin that he was having a dramatic response," said Benjamin.

But Susan Williams remembers the side effects that drug had on her husband. "It was like fire going through his veins. It burned . . . . It was just a constant seizure. They were trying to get it to stop, but it was his whole body. It was ugly and it was violent." she recalled.

On the evening of Nov. 22, Williams was given a general anesthetic to halt his epileptic seizures, according to hospital records. His blood pressure plummeted. At 5:45 the next morning, Marvin Williams died. His brothers, parents and wife were around him.

"I'm sure the death was drug-related," wrote Benjamin in his letter to NCI last April 23.

But Williams' 56-year-old mother, Martha Louise Williams, and his wife say they were never told anything about an "accident" or "overdose." Susan Williams recalls doctors telling her that her husband was responding to the drug, and that they had learned a great deal from his experience with it that would benefit others.

Keating declined to discuss the incident except to confirm Benjamin's account of it. He said he saw no reason to tell the family of the error. "I try to protect them from as much grief and difficulty as I can," said Keating. "It was not a cover-up."

The death certificate records the cause of death as "leukemia and terminal pneumonia." It is signed by Dr. Alberto Ayala, the M.D. Anderson pathologist who performed the autopsy on Williams' body. There is no mention of the drug on the death certificate.

It was not until Oct. 4, 1981 -- nearly two years after her husband's death -- that Susan Williams learned of the "overdose" in an interview with The Washington Post.

"Why couldn't they have told me? I always wondered why I didn't get an autopsy report . . . . You build up so much trust in the doctors," she said. "Why did he really die? That's what's going through my mind. I knew he was eventually going to die, but I didn't know when. If this made him die before he had to, I feel anger. I want to find out."

That same day, Williams telephoned Keating and asked him many of those questions.

"He told me how Marvin died . . . .He said it was an error, that Martin was not to be given another dose of Deoxycoformycin. He told me the drug had caused the seizures," Williams said, crying as she spoke.

"He told me I would finally get an autopsy report."

No Federal Approval

The M.D. Anderson Hospital in Houston is one of the nation's most prominent cancer centers. Its research and testing programs have a reputation for being among the most aggressive in the nation. In fact, three times this year alone, the National Cancer Institute has learned that M.D. Anderson was conducting human experimentation without authorization.

Deoxycoformycin is one of the "high priority" drugs listed by NCI director Vincent DeVita in 1979. But from the start, it has demonstrated very little action against cancer. "It has little potential as an antitumor agent" in animal systems, the NCI wrote in a 1979 report.

It is, however, a highly toxic compound. In early tests, Deoxycoformycin was implicated in eight patient deaths, according to NCI records.

A March 21, 1980, report filed with NCI described what happened to 11-year-old David Korn, who was given the experimental drug at the University of California in San Diego. Hours after receiving his third dose of the drug, the leukemic child developed "pulmonary edema fluid filled his lungs , followed by disorientation, confusion and hallucinations." He died within six hours, with the drug listed as a contributing cause of death.

Later in 1980, M.D. Anderson conducted experiments designed to find the highest dose of Deoxycoformycin that humans could tolerate. Twenty-two people were observed for their reactions to the drug. Many were nauseated and vomiting and suffered from inflamed membranes. Others experienced blood count drops, damage to the central nervous system, kidney failure and, one, like David Korn, had visual hallucinations.

Doctors at M.D. Anderson were eager to begin new experiments with Deoxycoformycin in combination with another experimental drug, ARA-A. They were so eager, in fact, that they started the experimentation without authorization from the NCI -- and despite specific instructions not to begin the testing.

On Jan. 6, 1981, according to NCI records, federal officials held a conference telephone call with 10 members of an NCI advisory group to assess M.D. Anderson's proposal to test the two experimental drugs together. The NCI officials decided not to approve the proposal for several reasons. They said that the proposed time schedule for administering the drugs "was not considered safe" and that the "dose of ARA-A was considered too high."

At about the same time, officials at the FDA raised other issues. FDA toxicologist Anthony Guarino wrote a detailed memo questioning the safety of trying two experimental drugs in humans before carefully testing them together in animals to observe side effects.

NCI asked M.D. Anderson to revise its proposal.

The hospital did not. Instead, it ignored the two federal agencies and went ahead with the experimentation of the two chemicals in humans, in what NCI said was an apparent violation of federal regulations.

On April 21, NCI learned of the unauthorized experiments. Daniel Hoth, head of the NCI experimental program, wrote to M.D. Anderson: "It has recently come to the attention of the National Cancer Institute that M.D. Anderson has instituted a trial of the combination of Deoxycoformycin and ARA-A. There is no approved protocol on file with the NCI . . . this situation is unacceptable."

The tests were stopped. Hoth ordered that the drug be returned to the NCI.

At a congressional hearing in June 1979, Dr. Emil J. Freireich, head of M.D. Anderson's therapeutic department, offered a general criticism of the federal government's caution concerning experimental drugs. Said Freireich: "One of the real tragedies of developing treatment is the conflict between safety and innovation and in the case of the Food and Drug Administration, we have now gone too far in the safety of drugs. We need some increased aggressiveness . . . ."

Less Than Partial Response

The origins of Bruceantin were a long way from this country's prominent cancer centers -- Sidney Farber Cancer Institute in Boston, M.D. Anderson, and the University of Kansas -- where testing of the drug began in 1978. Used in Ethiopia as an anticancer agent, the drug is extracted from the bark of a small flowering tropical tree in that country. It was one of the "high priority" drugs listed by DeVita in 1979.

The 117 patients who initially received the drug at three medical centers all had essentially the same result, according to an October 1980 NCI report. At the University of Kansas, there was no response among the 20 patients. At M.D. Anderson, there was "no objective" response among its 66 patients. At Sidney Farber, from among 31 patients, there was one partial response in a woman with cervical cancer.

The drug made some patients dizzy, nauseated, confused, disoriented, slurred their speech, made them lose their hair and vomit. In some cases, it made them stagger or partially paralyzed them.

Upon reviewing that record, NCI concluded that all studies should remain open and new ones should begin. "A study in cervical cancer is being considered since the Phase I results in this disease are encouraging," the report said.

The "encouraging" results were based on Dr. Mark Garnick's lone "partial response" reported in the cervical cancer patient. But Garnick said in an interview that he later reexamined the woman and concluded it was "less than a partial response." He decided not to continue testing the drug at Sidney Farber because it was "prohibitively toxic."

Still, at other medical centers, the studies of Bruceantin continue.

'Sour Grapes'

About the only thing the experts can agree on when it comes to "high dose Methotrexate" is that they disagree.

After more than a decade of human experiments involving thousands of people, the basic question of whether it is more useful against cancer than existing therapies remains in dispute. To examine what has happened with this drug is to look through a clear window at the drug development program in some of the larger research centers, a program where rivalry between researchers sometimes prolongs uncertainties. Methotrexate is one of the oldest and most commonly used anticancer drugs. It has been on the market since 1953. In the late 1960s, some doctors hypothesized that if side effects could be reduced, doses of Methotrexate could be increased and so too could the benefits of treatment. The idea was to boost the dose and try to "rescue" the otherwise endangered organs with a protective treatment.

But the side effects of the larger doses were often themselves "catastrophic," according to a joint report written by doctors at Stanford University in California and New York University Medical Center. Some patients' mouths erupted with sores so severe they could not eat; some vomited continuously and had to be hospitalized. Membranes were irritated and made painfully raw. There was kidney damage. Scores of people were subjected to "life-threatening toxicities," according to NCI's annual reports.

At least 39 deaths have been linked to the drug, according to those same records.

But researcher enthusiasm was not dampened. In 1978, some 106 doctors were experimenting with "high dose Methotrexate," and production and distribution of the drug topped $1 million a year, according to NCI records. At an NCI meeting on March 21 of that year, some doctors and administrators said that many of the "high-dose Methotrexate" experiments were "considered of minimal research value," according to NCI records.

In a 1980 paper entitled "Against High Dose Methotrexate: An Attempt to Rescue from Confusion," doctors at Stanford, New York University and Belgium's Institut Jules Bordet attacked what they saw as needless proliferation of tests.

"There is no proof for either enhanced antitumor selectivity or efficacy," they wrote. "Some clinicians, in their fervor to achieve good results and to make exciting theories work, have shed more smoke than light." They challenged the scientific reliability of studies that claimed the higher doses were curing cancer.

Then came the counter-attack: "So here we have M.D. Anderson, Sloan-Kettering, Sidney Farber, Roswell Park, Yale University -- all in favor. And who is against? . . . These are the best institutions we have in the United States and they can't all be wrong!" declared a speaker at a 1980 cancer conference in Frankfurt, Germany.

Among those who claim successes with the high dose program is Dr. Gerald Rosen of Memorial Sloan-Kettering. Rosen says that he can cure a large number of patients with osteogenic sarcoma, a rare bone disease, using the higher doses in conjunction with other treatments.

But hospitals cooperating in a study in the Southwest reported that none of 11 patients with the bone disease responded to the drug, according to an annual report last June.

Rosen maintains that those who do not get the same results either don't know how to use the drug or are intentionally changing the treatment in an effort to get NCI funding. "There's a lot of pressure on investigators to do their own thing to get grant money. If I publish my results, the people at Mayo, the people at Sidney Farber are not going to get funding to do what I do. They have to do something different," said Rosen.

"I've never seen so many people go more vehemently after negative results," said Rosen. "It's all a bunch of sour grapes."

Complete the Study

In some cases, doctors conclude from experiments that a drug has virtually no effect against cancer, but they decide to continue using the drug simply to complete the study they started. Such was the case with a drug called Anguidine, which had been derived from a parasitic plant fungus.

"Anguidine has been almost uniformly negative in a wide variety of human tumors," according to a February 1980 annual report by the Investigational Drug Branch of the National Cancer Institute. "Our plans are to continue the ongoing studies with this agent drug in order to provide enough data to complete the clinical panel."

Up to that date, studies in hospitals across the country found that, in 306 people given the drug, there was one complete response and 13 partial responses.

The experimental drug was considered useless as a treatment for almost all cancers. But doctors continued to ask people to take Anguidine despite the fact that it commonly caused nausea, vomiting, a drop in blood pressure, chills, diarrhea, fever, and a large range of central nervous symptom toxicities, including confusion, hallucinations and seizures.

In fact, in the February 1980 report, doctors noted the "consistent CNS central nervous system toxicity," suggesting that they try the drug against central nervous system tumors.

In a report published a year later, doctors concluded that the drug did not work against central nervous system tumors. By then, March 1981, 120 more people with several types of cancer were given the drug -- with no complete responses and eight partial responses.

Anguidine was one of the "high priority" drugs listed by DeVita in 1979.

'Prohibitively Toxic'

Some experimental drugs seem to come and go -- they are widely used for a few years, then dropped, then put back into experimental use. Methyl Gag is one such drug.

The chemical was first synthesized in 1958. Between 1968 and 1976 at least 262 men, women and children were injected with the substance. The drug was implicated in the deaths of at least 15 people, according to NCI records. The list of adverse reactions suffered by the rest seems without end: inflammation from their mouths to their intestines, blisters on their tongues, aching eyes, genitals swollen and painful, faces tingling with a prickling sensation, skin shedding from their swollen and tender hands and feet, dizziness, loss of appetite, nausea, vomiting, diarrhea, and boils and lesions on their skin.

So it came as no surprise when Hugh Davis Jr. of the NCI's Investigational Drug Branch wrote in the 1976 annual report that "interest in the drug is minimal." Davis went on: "Further study of this drug is not warranted and . . . no new investigations should be started." After all the years of testing, the verdict came in 1976: "prohibitively toxic."

The drug appeared headed for the warehouse. But that is not what happened. Seven days before the experimental drug file on Methyl Gag was slated to be closed, a University of Texas cancer doctor named William Knight said he filed an application to conduct one more study.

He proposed changing the earlier daily dose schedule to a weekly dose in the hope of reducing the cumulative effect of the drug. NCI approved the proposal and Knight began testing the drug in hospitals in the Southwest as part of a group study.

In its 1979 annual report to the Food and Drug Administration, the NCI reported the results of that study. Of 47 people on the experimental drug, "there are three complete responders," the NCI reported.

"Wrong," said Knight in a recent interview. "There were only two complete responses."

The NCI report also said that "this study provides strong hints of antitumor activity." NCI referred to the study's "exciting results."

That assessment is not shared by William Knight, the man who conducted the study. "I wasn't very excited," he said. "I think it was interesting. It depends on what one's definition of exciting is in cancer research. Obviously, you don't get excited about something like this. I don't think much of anything exciting in cancer. It's a slow stepwise process. I'll get excited when we cure something. That's the results of the study not my definition of exciting."

While the change from a daily to a weekly dose reduced the side effects, it did not eliminate them. The drug was implicated in the death of a man with liver cancer on the study, according to Knight. And 30 percent of the patients on the experiment suffered such side effects as mouth sores, loss of appetite, diarrhea, skin rashes, muscle aches, and inflammation of the membranes in the digestive tract.

Totally Bedridden

Cancer patients should have a life expectancy of at least two to three months to be admitted to a study of an experimental drug. In addition, in most experiments they should be ambulatory and have relatively normal body functions before the study begins.

Such was the case in Boston at the Sidney Farber Cancer Institute in 1978 when 51 patients were given an experimental drug called Maytansine.

"All patients were ambulatory at the start of drug therapy," Dr. Ronald Blum wrote in a report to the National Cancer Institute. "But in the two weeks following treatment, 30 percent of the patients spent up to 50 percent, and 18 percent spent more than 50 percent of their day in bed. One patient became totally bedridden, requiring assistance for daily care."

Hundreds of other patients across the country -- at M.D. Anderson in Houston, at the Mayo Clinic in Minnesota, at Mount Sinai Hospital in New York, and at the National Cancer Institute in Bethesda -- also were given this drug that had been derived from an East African shrub.

Numerous patients on Maytansine were reported to be suffering from "severe nausea and vomiting," "agitated depression," "alteration in sleep pattern," abdominal cramps, liver abnormalities, jaw pain, skin irritations such as burning, prickling feelings, diarrhea and constipation. Mayo Clinic reported a "possible cardiac toxicity."

Officials at the National Cancer Institute have concluded in the past month that, in the 293 people given Maytansine in the last four years, "no activity has been seen" in fighting cancer, according to an NCI document. There were no complete responses and only three partial responses, NCI said.

One of the partial responses occurred at Sidney Farber in Boston. However, the good news had a twist. Wrote Dr. Blum:

"A 59-year old man . . . had a greater than 50 percent objective regression of his cancer for three months. Unfortunately, this response was of neither symptomatic nor survival benefit to him, as he required hospitalization after each course because of severe drug toxicity. He died after three months from cachexia deterioration , while still demonstrating an objective response."

According to an NCI document, the National Cancer Institute plans to stop use of Maytansine -- which was also on DeVita's "high priority" list -- in the near future.

Used in Combination

A close relative to several drugs already commercially available, Chlorozotocin was developed in the hope that it would be less toxic than many other experimental anticancer drugs. It was first put to the test in 1975, and the results were less than overwhelming. The November 1979 report on the drug noted that, of 460 patients who had taken it, one had a "complete response." Another 27 had "partial responses."

Many patients experienced nausea, vomiting, diarrhea, weakness, disorders of the central nervous system, including agitation and anxiety, and radical changes in their blood.

Although there was only one complete response, the NCI concluded in its 1979 report that "early reports are indicating activity in several tumor types." By March 1981, the most recent evaluation of the drug, some 1,212 patients had been given the drug. Ten had shown a complete response, with another 52 showing "partial response."

Having been tested against a whole spectrum of cancers, researchers concluded last March that Chlorozotocin showed activity against only one type of cancer -- melanoma. But, the NCI said, the drug appears to be no more effective than existing treatments. There is no known cure for melanoma.

Experiments with Chlorozotocin continue, in combination with other drugs, in the hope it will be less toxic than other therapies.

Chlorozotocin was one of the "high priority" drugs listed by DeVita in 1979.

Footnote: The sole complete response to Chlorozotocin recorded among the 460 patients initially reported was in fact not a complete response at all. The physician, Dr. Robert Talley, says he was out of town at the time of his patient's examination and the examining physician failed to note a cancerous node in his patient's right arm. Talley says he expects the cancer to have returned by his patient's next visit.

'Strange Syndrome'

A drug called 5-Azacytidine (5-AZA) has shown effectiveness in treating some forms of leukemia, according to a March 1981 NCI report. It is one of NCI's more widely used experimental drugs, and more than 1,000 people have received it. But some doctors heading the research efforts have raised serious questions about 5-AZA.

In August 1975, when large-scale human experimentation was to begin, NCI warned doctors of serious reactions in early human experiments, including: liver damage, sometimes leading to coma and death; severe blood count drops, sometimes lethal; hypotention drop in blood pressure which was at times fatal; and severe nausea and vomiting.

The NCI also wrote: "A strange neurological syndrome has been described with myalgias muscle pain and lethargy sometimes progressing to coma. No reason for this toxicity has been established."

The same 1975 report also described the agony that most patients suffered, severe prolonged nausea and vomiting, after receiving an injection of 5-AZA. Dr. Charles Moertel of the Mayo Clinic in Minnesota wrote: "The severity of nausea and vomiting induced by 5-Azacytidine seriously compromises any hope of clinical usefulness."

The NCI decided to proceed with large-scale testing of the drug believing the potential benefit of the drug outweighed the risk.

Last March, doctors working cooperatively in seven hospitals reported the results of one of the largest studies with 5-AZA. Of 170 patients with leukemia entered in the study, there were nine complete responses and two partial responses. The median length of the nine complete responses was 65 days.

Dr. J.H. Saiki, of the University of New Mexico School of Medicine, wrote the report of the study. The toxicities he found reflected the early warnings published in 1975. "The severe nausea and vomiting associated with the drug therapy significantly impairs its usefulness," he wrote. He also reported that "several patients developed irreversible myelosuppression bone marrow destruction hypotension, and a syndrome of myalgia, fever, lethargy and coma which appeared to be secondary to related to 5-Azacytidine."

Although he found the therapeutic results promising, Saiki concluded that 5-AZA was not valuable enough to continue using alone in treating leukemia patients. He suggested that further experimental studies be done using 5-AZA in combination with other drugs.

"Further studies of 5-Azacytidine as a single agent do not appear to be warranted," he wrote.