• The management of clinical trials
  
• Investigators and their staff
  
• Institutional Review Boards
  
• Multicenter trials
  
• Data Safety and Monitoring Boards
  

Food and Drug Administration

The path to approval for a new cancer therapy is a complex one. What follows is a highly abbreviated summary. For more particulars, see CDER's web site at: www.fda.gov/cder and especially the "CDER Handbook" at: www.fda.gov/cder/handbook.

Before clinical trials can even begin, the sponsor--an individual, partnership, corporation, government agency, manufacturer, or scientific institution--must submit an Investigational New Drug (IND) application to CDER. CDER reviews the preclinical data about the drug's safety and effectiveness. It reviews information related to how the drug is manufactured, it reviews the proposed protocols for the clinical trial, and it reviews the qualifications of the clinical investigators. If CDER judges all these factors to be acceptable, it will permit clinical trials to proceed. Technically, though, what CDER grants when it approves an IND is an exemption to the Federal statute that prohibits unapproved drugs from being shipped across state lines.

With the IND in hand, the sponsor conducts clinical trials. Once Phase III clinical trials are completed successfully, the sponsor submits a New Drug Application (NDA) to CDER. NDAs are massive filings, occupying many thousands of pages containing every detail about the scientific and clinical aspects of the new treatment and the clinical trials that establish its safety and effectiveness. Records of each individual participant in all the clinical trials are included. CDER reviews every aspect of the NDA, and the application is also reviewed by an FDA advisory committee--a group of outside experts. If the NDA passes all those hurdles, and the advisory committee recommends approval, the FDA grants the manufacturer the right to market the drug for certain conditions.

In 1996, the FDA undertook several initiatives designed to improve patient access to new cancer therapies. The most important of these initiatives allowed sponsors to demonstrate a new treatment's effectiveness using a "surrogate marker." Previously, sponsors had to demonstrate actual increases in survival time for a new therapy to be approved. Since it often takes quite a long time to accumulate survival-time statistics, the approval process often moved slowly. Beginning in 1996 the FDA began approving treatments when the sponsor merely demonstrated tumor shrinkage as an early indicator of the treatment's effectiveness, thus speeding the approval process.

The management of clinical trials

One common scenario occurs when clinical trials are initiated by individual investigators or small groups of investigators at academic medical centers. They obtain funding in the form of grants from the NCI or from pharmaceutical companies. Often investigators band together into what are known as "cooperative groups," which manage multicenter trials (see below).

These cooperative groups differ in structure and research focus. Some, such as the Southwest Oncology Group, bring together investigators in a certain geographical region. Others, such as the Radiation Therapy Oncology Group, study a specific type of cancer therapy. Others, such as the Pediatric Oncology group, concentrate on a certain type of patient. Still others, such as the Gynecologic Oncology Group, focus on a group of related cancers.

Dr. Richard A. Gams, the Chief Scientific Officer at Prologue Research International, Inc., is an oncologist who has been associated with several academic institutions and several pharmaceutical companies. He explains how management differs when a trial is being done for FDA submission:

In a clinical trial done in an academic medical center, generally the NCI or a company may simply turn over a therapeutic substance to academic investigators, who will be fully responsible for the design and conduct of the trial. They will do their own data management, they will do their own reporting. This will result in a publication in a scientific journal. In a trial designed for FDA submission, however, the sponsor will usually write the protocol, recruit an investigator to participate in the protocol, file the protocol with the FDA, actually send its own personnel out to the site to monitor the conduct of the trial, bring the data back to the company, do all the final analyses, help the investigator to publish the study, and submit the final study reports to the FDA.

I think in the 1980s we moved from what I would call the "grant era" of doing clinical trials-where someone would give an investigator money to do what the investigator wanted to do--to the "contract era"-where we would give investigators money to do what we wanted them to do. The FDA began taking a much more critical look at the data. Pharmaceutical companies began recognizing that university investigators often were not as meticulous in meeting all the requirements of the regulatory environment. Pharmaceutical companies ultimately recognized that rather than depend on the investigators to initiate the studies, and rather than depending on the cooperative groups to identify the investigators, they began to control the trial completely from within the company. They had to develop a medical department that consisted of physicians, clinical research associates who would actually go out and visit the sites and monitor the studies, information systems departments, biostatisticians, etc.

Whether a trial is sponsored by a drug company or by the NCI, in reality the conduct of the trial is no different. It has to be the same rigorously controlled environment, otherwise you can never interpret the data at the end of the day. The only things that are a little different are the motivations to a certain extent, what kind of questions different groups are answering. Drug companies usually have very focused questions regarding certain stages of patients and what the efficacy end points are. NCI or other government-sponsored trials are sometimes a little more open-ended. Drug companies need to have very focused questions, because clinical trials are enormously expensive. They have to choose carefully and wisely where their likely efficacy will be seen in the quickest period of time.

At first the CROs were highly specialized. One would provide biostatistical advice, helping the pharmaceutical companies analyze the data from clinical trials. Another would provide people called clinical research associates who would go to the academic institution and monitor the conduct of the trial. Still another would specialize in regulatory affairs, helping the company compile its submissions to the FDA.

Eventually, full-service CROs began to emerge. When a pharmaceutical company had a drug that was ready for clinical trials, it would hire a single CRO to manage the trials from start to finish. The CRO would write the protocol, it would recruit researchers at a number of hospitals, it would help recruit patients, it would oversee the conduct of the trial, it would collect and analyze the data, and it would prepare the FDA submission.

Many cancer clinical trials are conducted this way today. Interestingly, few patients realize this, since the CRO is often invisible. Typically, your direct contact will be with the individual investigator, and if you ask you'll be told which pharmaceutical company is sponsoring the trial. Unless you asked, however, you probably would never realize that there was an entirely separate entity running all the day-to-day details of the trial. This is not necessarily a bad thing. For the everyday details of a clinical trial, you'll be dealing directly with the investigator and his staff. In the event of serious problems that can't be dealt with at that level, you'll be dealing with the institutional review board (see below). For problems that can't be resolved at the IRB level, you can contact the FDA. There are circumstances in which you may wish to deal with the pharmaceutical company sponsoring the trial. Although it might be nice to know about any CROs that are involved in the trial, from a patient's point of view, it will rarely be necessary.

As mentioned above, pharmaceutical companies don't sponsor all trials. Other clinical trials are sponsored by the NCI and conducted by NCI-funded "extramural" (outside) institutions--often major medical centers. But not all trials are conducted by physicians at large hospitals. In 1983 NCI established the Community Clinical Oncology Program (CCOP), a way for physicians who are not located at academic medical centers to participate in clinical trials. Physicians participating in the CCOP will establish an affiliation with either a cooperative group or a major cancer center, and they enroll their patients in clinical trials whose protocols were developed by their affiliates. The advantage of the CCOP is that it brings the possibility of participating in clinical trials to patients who may be a long way from a major cancer center. If you're not being treated at a major cancer center, you should ask your oncologist whether he's part of the CCOP.

Still other NCI clinical trials are "intramural"--conducted at the NCI Clinical Center in Bethesda, Maryland. According to Dale Shoemaker, chief of the regulatory affairs branch of NCI's Cancer Therapy Evaluation Program, the NCI often conducts research on agents that may not currently have a pharmaceutical collaborator. If clinical trials reveal this agent to be promising in cancer treatment, the NCI will advertise for a pharmaceutical company able to shepherd the drug through the FDA's approval process, and able also to market the drug once it's approved.

Investigators and their staff

The PI may or may not be the individual directly providing your medical care. Sometimes two or more physicians at an institution have equal responsibility for the conduct of a trial. When that's the case, they're referred to as co-PIs. Other times the person providing your medical care will be a subordinate of the PI.

You'll also be dealing with several people who are not physicians. Many PIs employ a physician assistant or a nurse (sometimes called a research nurse or a nurse coordinator) to handle most of the administrative details involved in clinical trials. Gregory T. David works for Dr. David Jablons at UCSF in this capacity. As thoracic surgery nurse coordinator, Gregory's would be the first voice you heard if you phoned regarding one of the trials Dr. Jablons was running. During that call Gregory would ask you questions to determine whether you were eligible for the trial, and he'd arrange your initial and subsequent visits. He'd also work with you and your insurance company to obtain reimbursement, and he'd handle a thousand other administrative details.

Another person you're likely to encounter is called a clinical research associate (CRA). Employed sometimes by the PI, and sometimes by the pharmaceutical company or contract research organization, it's the CRA's job to manage the mountain of paperwork generated in each clinical trial and to ensure that every aspect of the protocol is followed to the letter and clearly documented.

Institutional Review Boards

The law is very specific about the makeup of the IRB. The IRB must have at least five members of varying backgrounds who are knowledgeable about human research. This committee must not be composed of all men or all women, and must be sufficiently diverse--by race, gender, cultural background, and sensitivity to community attitudes--to "promote respect for its advice and counsel in safeguarding the rights and welfare of human subjects." Every IRB must include at least one member who is not affiliated with the institution in any way. This person is generally referred to as the public member. And every IRB must include both scientists and non-scientists.

Although only five members are required, most major research institutions have far more IRB members than that. In part this reflects a desire to obtain a wide range of views, but it's also an effort to divide the large workload among a greater number of people. For example, UCSF's IRB typically has between thirteen and eighteen members, according to Sharon K. Friend, the university's IRB administrator. The UCSF Committee for Human Research is usually made up of:

• One to three oncologists
• An AIDS specialist
• A pediatrician
• A radiologist or nuclear-medicine specialist
• A pharmacist who specializes in investigational new drugs
• A clinical nurse specialist
• A behavioral scientist
• A psychiatrist
• An anesthesiologist
• An internist
• A surgeon or obstetrician/gynecologist
• An epidemiologist or biostatistician
• The public member

Federal law is also very specific about the IRB's job. It gives the IRB authority to approve, require modification to, or disapprove all research involving human subjects. Moreover, the IRB is charged with overseeing informed consent, and with conducting annual reviews of all ongoing human research.

In practice, every investigator who wishes to conduct a clinical trial must submit the detailed protocol, as well as the proposed informed consent documents, to the IRB for approval. The IRB will conduct a close examination of both the scientific validity and the patient-protection aspects of the trial. Some institutions have a separate committee, often called the Protocol Review Committee, that concentrates on the scientific aspects of the protocol.

Dr. Victor Santana notes:

In the past year we have rejected one protocol out of twenty-five new ones submitted for review. That specific proposal was something very risky, and there was no direct benefit to the patient. I would say that 85% of protocols require some modification, either in the protocol document itself or in the informed consent, upon IRB review.

There's a lot of attention paid to the who, when, where, why, and how of the consent. We wouldn't want to see somebody in labor asked to give consent. There's a lot of discussion about when in terms of surgery a patient would get talked to about a project. Is it okay the morning of surgery? Who is the best person to be talking to the patient. What if they're trying to recruit people through medical records? Will they be sending out letters or making phone calls or putting up notices? We'd want to see all that. If there's a national ad campaign, we'd like to see the video or the text. The main thing we look at with ads and recruitment materials is that they're not over-promising benefits.

We are the one place that's one-hundred percent for the subjects. If somebody calls us with a problem, then we're on the phone right away with the investigator, and they take our calls very seriously.

Our IRB is relatively slow and relatively difficult. It can take three months. It can take six months. It can take a year. Usually the stumbling blocks are not major philosophical things. They're detail oriented, and they may be driven by semantics or personal agendas. I can't tell you the amount of Sturm und Drang that's generated over the wording of consents. It's a bunch of crap. I think there should be one standardized, nationwide consent, to which you'd add the individual experimental details. It's a huge problem. These patients aren't waiting around. They're dying quickly. It's very frustrating from the perspective of trying to be an effective machine for clinical trials, to attract the best, the hottest, the most interesting science, and to bring it to the clinical arena.

One might worry that a freestanding IRB would be likely to rubberstamp any protocol handed to it by the company that's also paying its bills. But Dr. Richard A. Gams, chief scientific officer of Prologue Research International, Inc., an Ohio-based CRO that specializes in cancer trials, denies that that's a problem. He notes that freestanding IRBs are subject to the same Federal laws as are institutional IRBs, and that in any case institutional IRBs have similar potential conflicts of interest. After all, most members of an institutional IRB are employed by the institution, and they have an interest in approving protocols so that the institution can garner the funding and the prestige that come with conducting clinical research. Despite that, notes Dr. Gams, institutional IRBs are anything but pushovers, and he maintains that this is true of freestanding IRBs as well.

Multicenter trials

These are known as multicenter trials, and they have several advantages over single-center trials. As the saying goes, many hands make light work, and multicenter trials decrease the burden on any single institution. Instead of having to find and treat hundreds of patients, a single institution may only have to deal with a few dozen. With several institutions in different regions of the country all searching for participants, patients can be enrolled in the trial more quickly, and the trial can be concluded and its results determined more promptly. Since patients will be recruited from widely separated geographical areas, the treatment will be given to a group of people that's not overly homogeneous, and is more representative of the population at large. From the patient's perspective, multicenter trials increase the likelihood that an innovative treatment may be available locally, so travel may present less of a problem.

But multicenter trials present their own set of problems. When you have a number of different physicians with different levels of expertise and different styles all treating participants in the same clinical trial, there's a danger that the data at different sites may not be comparable. Dr. Jablons explains:

The problem in multiple sites is you have to make sure there's good quality control. It's not that the doctors aren't good at all these academic institutions, but for example, if you're not trained to stage patients the appropriate way, earlier stage patients are thrown into the mix, and that skews the data.

It's unclear how much of a problem this possibility truly presents. IRB members insist that they examine protocols for multicenter trials as carefully as trials that originate within their own institutions, and they say they're not afraid to demand changes when necessary. Dr. Santana gives this example:

I remember one protocol that was submitted after already being activated at another hospital. The IRB felt very strongly that there was a flaw in the study design, and we said, "We will not approve this study, for these specific reasons. Go back to the cooperative-group committee and raise these concerns." Lo and behold, the problem was recognized and fixed, and the protocol came back to us. It's naive to assume that if it has been approved by other IRBs, it's flawless.

Data Safety and Monitoring Boards

This is called the Data Safety and Monitoring Board (DSMB), also referred to as the Data Monitoring Committee (DMC). DSMBs are established at the level of the cooperative group. DSMB members are independent experts who are selected for their experience, objectivity, absence of conflicts of interest, and knowledge of clinical trials.

It's the job of the DSMB to examine the interim data emerging from an ongoing clinical trial, and to determine whether the trial needs to be changed or terminated. For example, if the treatment under study seems to be causing too many adverse reactions, the DSMB might recommend that the dose be lowered or that the trial be ended. They can also recommend that the trial be ended if the experimental treatment is so effective that it would be unethical to continue giving some patients the standard treatment. In that case all patients receiving the standard treatment would be given the opportunity to get the experimental treatment.

But DSMBs have their critics.¹ Unlike IRBs, DSMBs are not mandated by Federal law and are not currently subject to Federal regulation. These critics note that while the responsibility for protecting subjects is assigned to the IRB, the exercise of that responsibility often falls to the DSMB, which is not accountable to and does not interact with any IRB. The critics reason that the DSMB's responsibilities and procedures ought to be codified. While the NCI does have specific written policies governing the conduct of DSMBs, these policies may not have the force of law.

Notes:

1. Charles R. McCarthy, "Challenges to IRBs in the Coming Decades," in Vanderpool, The Ethics of Research, 127-44.