We looked briefly at why drug studies came into being; now let’s look at how a drug is developed, from test tube to your tissues.

Every government approved drug goes through the same sequence of testing anywhere in the world. In the US, this is done under the supervision of the FDA, and is conducted in accordance with international standards–mostly. (The US has bowed out of signing the last Declaration of Helsinki; more on that when we talk about ethics later). Each phase of testing is intended to capture data about the drug’s efficacy and safety, adding more patients as more experience is gained in each step. A similar process of testing occurs for medical devices and “biologics,” or vaccines, blood products, and gene therapies; we will just refer to “drugs” here for all.

Stages of Drug Development: Preclinical

The first phase of study is preclinical, and is the testing that occurs in computer modeling, test tubes (in vitro) or in animals. Thousands of compounds might be screened in the lab in search for a potential winning candidate compound. Next, the winning agent is given to animals in order to study the drug’s action and metabolism and to look for obvious toxicities before it is given to people. No one likes to hurt animals–I still have nightmares about having had to sacrifice and dissect mice on Thanksgiving in 1973. But animal studies are needed to protect people, by identifying toxicities and weeding out drugs before testing on humans. And I’m gratified to see that the testing I helped with while in college have helped lead to treatments for a lung disease, anemia, and chemotherapy related side effects now.

These preclinical studies take a very long (5-7 years) and arduous path, with only a small percentage of drug candidates surviving this phase.

Estimates of the number of drugs that successfully pass the preclinical hurdle vary substantially, from only 1 of 1,000 (Tufts CSDD) to 10 percent (NIH). “Drug developers colloquially call this (period) the ‘Valley of Death.’”

Investigational Medicines Reach People

Phase 1: Safety and Pharmacology

In first-in-human studies, or Phase 1 trials, 20-100 volunteers are given small amounts of the study compound in order to test its

Researchers are trying to find out as much as possible as to how and where the drug works, on a basic physiologic level. For example, some antibiotics, like aminoglycosides, work in a dose-dependent fashion, called concentration dependent killing. So these drugs tend to be administered in less frequent, high doses. Others, such as cephalosporins and penicillins, kill bacteria in a time-dependent fashion, based on the length of exposure to the drug. With these drugs, you don’t have to give high, potentially toxic doses to be effective.

Phase 1 (the first time an experimental medicine is given to humans) studies are often done in real patients in cancer trials, with patients volunteering out of a sense of altruism or desperation. Otherwise, Phase 1 trials are commonly conducted in healthy young people who receive no benefit from their participation, except financial. Understandably, these early trials are perhaps the most often criticized as exploiting poor people and using humans as “guinea pigs.”

As you might expect, Phase 1 trials are the riskiest, so are not for the faint of heart. You just don’t know what might happen. Reassuringly, a review of oncology trials showed no increase in deaths compared to other, not first-in-human drugs, and noted that some participants received significant benefit. Most clinical trials are thoughtfully planned out and conducted with close attention to detail. A notorious exception to this is the TeGenero TGN1412 trial, in which six young, healthy men became critically ill, developing multisystem organ failure. This tragedy will be a classic example of what not to do in conducting a trial.

To try to reduce the risk in Phase 1 trials, dosing is started at less than 1/10th of the human equivalent dose seen in the animal studies. “Microdosing,” or Phase 0 trials, are now sometimes substituted for the traditional Phase 1 testing, using tiny, incrementally larger doses.

Early phase 1 and 2 studies also look at factors that might affect absorption, such as different product formulations, or how taking the drug with food or antacids might change absorption. That’s why some prescription labels say, “Take on an empty stomach,” or “Take with meals.” This can get very complicated–for example, antacids or dairy products bind to some antibiotics, such as quinolones or tetracycline, preventing your body from absorbing your medicine. Yet grapefruit juice increases the blood level of many HIV medications, causing toxic levels!

A drug’s metabolism and excretion has to be studied to make dose adjustments for patients with impaired liver or kidney function. Cardiac tests are now required to look for possible life-threatening arrhythmias. Finally, drug interaction studies are done to look for possible problems. Such interactions notoriously occur with the blood thinner, Coumadin (warfarin), for example. Less well-known are serious interactions with antibiotics and even antihistamines. Seldane (terfenadine) and Hismanal (astemizole), for example, were withdrawn from the market because of such problems.

Knowing how a drug is metabolized can help predict whether there are likely to be serious interaction problems and will be important later in drug labeling and prescribing information…but uncertainty remains.

Phase 2: Dosing and Efficacy

In phase 2, the “dose-finding” phase, the drug company (sponsor) determines efficacy for the drug’s intended use and tries to find the best dose for the target indication. Initially, patients are generally not very ill, (e.g., with uncomplicated pneumonia or urinary tract infection), nor do they have many other illnesses or medications that could lead to confounding and confusing results. A bit later, when the untested medicine is first given to very sick folks, was always a particularly spooky time for me. Many of these patients in my studies of pneumonia and sepsis, for example, were already pretty frail and quite ill. The responsibility of trying something new weighed heavily.

Phase 3: Efficacy–Will the Experimental Medicine Win Approval?

Phase 3 broadens the population that receives the new drug, including more real-world patients who do have other medical problems (underlying diseases). In phase 3, patients receive either the new study medication or one that is already on the market. Depending on the illness under study, one group may receive a placebo (a fake, phony medicine–the proverbial “sugar pill”) or sham (fake) surgery.

Phase 3 is the definitive phase before the sponsor submits a New Drug Application (NDA) to the FDA. An NDA claims the drug’s effectiveness in treating a particular illness. Phase 3 studies are often large (thousands of patients) and multicentered (conducted at multiple sites, usually covering a wide geographic area) and are considered primary efficacy studies, or pivotal trials in demonstrating a drug’s efficacy. Generally, two successful phase 3 trials are required in order to obtain approval from the FDA (or similar international regulatory agency such as the European Union’s EMEA). The rules are less stringent for oncology trials, where one successful efficacy trial is required.

One of the most commonly misunderstood issues about clinical trials relates to placebos. Sometimes, in addition to searching for the optimal dose, a placebo (or no treatment) arm is used as well. You might see this, for example, in studies assessing the value of adding a vitamin or symptomatic treatment to a patient’s regimen. It is important to emphasize that placebos are never given to patients who are seriously ill if an alternative therapy is available. To do so would not only be unethical; it is also illegal. Use of placebos in clinical trials is seriously frowned upon by the Declaration of Helsinki; again, the US is not a signatory to this international agreement.

Because phase 3 trials are so important to the drug’s (and the company’s) success, outcome and safety data are often monitored by an independent Data Safety Monitoring Board, especially if members of the sponsor’s team are blinded. [The term "blinded" means that the sites participating in the trial do not know which patients are getting the new compound and which are getting the standard regimen or the placebo. The purpose of this is to prevent the investigators from unintentionally biasing the study results.] The DSMB may occasionally recommend changes during a trial. It can also halt the trial at any time because of safety concerns or because its analysis of outcomes shows that one treatment group is faring significantly better than the other, and therefore it would be unethical to continue the trial.

For example, the Tenofovir arm of the VOICE HIV trial was dropped last month after the National Institute of Allergy and Infectious Diseases (NIAID)’s DSMB found they would not be able to determine efficacy of the drug in preventing HIV, compared to placebo. The four other trial arms are continuing, as no concerns were raised.

In contrast, the higher dosage arms of Elan’s ELND005 Phase 2 study for an Alzheimer’s drug were halted after the DSMB found increased deaths in the groups receiving the higher dosages. The lowest dose group was allowed to continue on the trial.

The race is on!

There is an enormous market advantage (read, profit) to the company that has the first effective drug for an ailment. That first drug generally becomes the standard of care to which newer drugs are compared.

When the Investigational New Drug (IND) application is filed, the drug is patented for 20 years; the clock is ticking. This largely explains the marked competitiveness among companies to develop the first drug for an indication.

Phase 4, or Post-Approval Studies

If the drug or device survives these initial phases, it can then win FDA or European Medicines Agency (EMA) approval. Post-approval studies continue as Phase 4 trials, which often seem driven by marketing considerations, rather than intellectual curiosity. These trials compare the new drug, already approved by the FDA, to one that is viewed as the major competitor for the same indication. In phase 4 trials, further safety data are gathered, sometimes at the FDA’s insistence or as a condition of approval of the NDA. For example, as a condition of approval for the blockbuster drug Xigris for treating early severe sepsis, Eli Lilly and Company was required to continue to study its drug, postapproval, in many thousands more patients who are less critically ill.

The drug maker tries to expand the approved uses to other indications at this time. Phase 4, or postmarketing studies, can also lead to a change in a drug’s status from prescription only to over the counter. And phase 4 studies may target new groups of different ages, sexes, or ethnicities.

But are these trials safe?

There are always going to be unfortunate surprises when testing anything new, but there are safeguards in place–and they work, most of the time. When they don’t, you see sensational headlines like this:

Drug trial creates ‘Elephant Man’

Except in homeopathy, there is a generally a correlation between efficacy and toxicity. You might picture the balance between safety and efficacy as being weighed on the on the scales of justice. Clinical trials aim to find the “sweet spot” of balance between the two.

Dosing: Safety vs. Efficacy
Chlorine concentration in H20

For example, let’s look at chlorinating water. At too low concentrations, water would be unsafe to drink because of the various bacteria and viruses in it. After filtering and chlorinating, it will become potable. But at high concentrations, it might taste like a swimming pool and, in larger quantities, make you ill. In very concentrated forms, chlorine compounds are quite caustic and toxic.

Why aren’t all the bad effects found before marketing?

It all comes down to a numbers game. Even after a drug is marketed, surveillance for safety continues. Occasionally, this postmarketing surveillance uncovers serious side effects that have not been previously recognized, resulting in the drug either being removed from the market or warnings being added, and much finger pointing among different oversight agencies. An example was the previous illustration of the Thalidomide tragedy in 1957; kids are again being born with phocomelia as the drug is again being used, especially overseas.

Sculpture by Marc Quinn; photo by Louise Joly; acc 10/14/07. CCAtr

The most egregious recent example was that of Ketek (telithromycin). Conscientious FDA reviewers, as Dr. David Ross, were allegedly threatened by then-FDA Commissioner Andrew von Eschenbach and were told not to express their opposition to Ketek’s approval or they would be “traded from the team.” One hopes such incidents are anomalies.

Because trials involve only small numbers of patients, often in controlled (rather than real world) conditions, post-marketing discovery of adverse effects of drugs is inevitable and unavoidable, despite careful reviews at each level to assess efficacy and safety. For example, for medications for chronic conditions, the International Conference on Harmonisation guidelines specify the number of subjects required for approval at approximately 1,500. Most adverse events occur within the first 6 months of exposure, so you need 300-600 patients treated for at least that length of time to detect events occurring at a frequency of 0.5-5 percent. To detect AEs with a cumulative 1-year incidence of 3 percent or less, more than 100 patients treated for more than a year are required. Since most trials are shorter in duration or involve fewer patients, side effects and toxicities may go undetected and become apparent only after the drug is in wide use.

Safety withdrawals occur for approximately 3 percent of drugs and are not lower in the United States than in Europe, where the approvals are speedier.

It shocked me to learn the magnitude of unnecessary drug exposures. According to Dr. Alastair J. J. Wood, assistant vice chancellor for research at the Vanderbilt University Medical Center, “First, a staggering 19.8 million patients (almost 10% of the United States population) were estimated to have been exposed to just 5 of the 10 drugs withdrawn in the past 10 years. Second, none of the drugs was indicated for a life-threatening condition nor, in many cases, were they the only drugs available for that indication.”

The FDA is in the position of being damned, on the one hand, for not subjecting drug candidates to closer scrutiny for safety and, on the other, for deaths and morbidities from delays in approval. Daniel B. Klein and Alexander Tabarrok give a fascinating review of this dilemma, concluding that far more lives are lost by the delays in new product releases than are saved by the added safety observations.

So, we’ve seen that clinical trials go through a number of phases of testing over a period of years, before the drug that was being tested is approved by government regulatory agencies for marketing to the public. Some urge less oversight and review, others a more measured pace, valuing safety above all. Despite care at all levels, unexpected side effects will inevitably found once a drug or device is in widespread use. That is the price of medical innovation.


This post is adapted from my book, Conducting Clinical Research: A Practical Guide for Physicians, Nurses, Study Coordinators, and Investigators.