Sci. Aging Knowl. Environ., 16 June 2004
Vol. 2004, Issue 24, p. oa1
[DOI: 10.1126/sageke.2004.24.oa1]


On Trial

Implicit in any basic research investigation of aging is the hope that its findings will apply to people. But testing theories from the lab in human populations presents unique challenges

R. John Davenport

Abstract: As researchers uncover basic mechanisms that underlie aging, they hope that lessons learned from mice, worms, flies, and other lab denizens will apply to people. Making that leap requires a new set of scientific approaches, from observational studies to the most rigorous of tests, the double-blind, randomized, placebo-controlled clinical trial. Navigating this unique scientific terrain presents new obstacles for basic-science researchers but can provide the ultimate payoff.

Introduction Back to Top

Creating an age-resistant fly might inspire researchers to pop open the champagne, but translating that work into a health-extending therapy for humans would really give them a reason to party. Underlying any laboratory study on the basic mechanisms of aging is the hope that the rules learned in model organisms will apply to humans. Such insights might help investigators design interventions and therapies that assuage age-related disease or stall the aging process itself. But testing drugs in humans is a different scientific universe from manipulating lab animals or scrutinizing cells. For the basic-science researcher, conducting clinical trials presents substantial challenges: new bureaucratic hoops, a slower pace, recruitment of the right subjects, and trickier data interpretation. Negotiating these hurdles, however, offers the potential to translate lab findings into treatments that improve human life.

View larger version (12K):
[in this window]
[in a new window]
People power. Each type of human study has its benefits and drawbacks.


The Human Touch Back to Top

Clinical trials are the gold standard for assessing the safety and effectiveness of drugs or other treatments. But before clinical trials even begin, investigators often get clues about how medications, diet, or other interventions affect health by conducting so-called observational studies. These enterprises can be either retrospective, in which case researchers assess effects by collecting historical data, or prospective, in which they track a group of people going forward in time. In a retrospective study, for instance, investigators might quiz patients with heart disease about their past eating and exercise habits and then compare their answers to those of people without the affliction; behaviors that are more common in one group than in the other might indicate risk factors or identify potential treatments. A prospective study might track people over months or years, cataloging their behavior and noting which individuals develop heart disease or dementia. In either situation, researchers might notice that people with heart problems tend to gobble French fries daily or that those whose brains remain sharp do crossword puzzles regularly.

These research tactics help researchers develop hypotheses, but they have limitations. Participants don't always produce reliable accounts of their behavior. In addition, a connection between two factors might be a red herring; for instance, a study could reveal that people who down multiple martinis each day are likely to get lung cancer--but only because people who drink are also more likely to smoke. Moreover, in observational studies researchers don't control how subjects behave, making it difficult to pinpoint a single factor responsible for a difference between groups. For instance, scientists have found that people who take vitamin E resist heart disease. But such individuals might be more likely to eat healthily, exercise regularly, and visit a doctor than do patients who don't consume supplements, thus clouding conclusions about whether the vitamin helps fight disease. In addition, participants might take different amounts of a drug or work out to different degrees.

So researchers often need to go beyond observational studies to provide convincing evidence that a treatment or a lifestyle choice alters the course of a disease, says oncologist Norman (Ned) Sharpless of the University of North Carolina, Chapel Hill. Clinical trials trump observational studies in several ways. Researchers determine exactly how much and what treatments study subjects receive. In addition, subjects are placed randomly into groups that either do or do not undergo the test intervention. This randomization ensures that other factors--such as age, weight, socioeconomic status, and gender--are represented equally in the two groups, effectively canceling out their effects.

The strongest trials are placebo-controlled: Some volunteers receive the new regimen and others get an inactive substitute. This design enables researchers to attribute benefits to the test compound rather than to the psychological value of receiving a treatment, which can be dramatic. And the ultimate trials are double-blind, meaning that neither physician nor subject knows whether the subject is receiving the drug or the placebo, further removing bias that might skew results. For instance, doctors could inadvertently downplay side effects in patients who they know are receiving a promising new drug. "Your doctor might say, 'Oh, buck up, you'll probably be better in a few months," says Sharpless. "Whereas if it were blinded, [the doctor would] just say, 'Either the drug or the placebo could be causing your problem, and there's no way to know which. But we'll write it down because I'm sure it's going to be important.' " Blinding isn't feasible in all trials, however; for instance, cancer patients and their physicians would know whether chemotherapy or surgery was delivered. In each case, researchers strive to design studies that eliminate as many complicating factors as possible.

In short, observational studies provide hints that a treatment might be effective, but clinical trials test that idea. If they pan out, they help researchers determine how much benefit a drug provides and zero in on the most effective and least toxic dose.

View larger version (13K):
[in this window]
[in a new window]
Proving grounds. Clinical trials provide the ultimate proof that a treatment is safe and effective.

Clinical trials don't always back up observational studies, and sometimes they spur doctors and patients to rethink decades-old conventional medical wisdom. For instance, recent placebo-controlled clinical trials failed to back up observational data that hormone replacement therapy (HRT) prevents heart disease and improves cognition. Instead, they showed that it might exacerbate circulatory problems and spur cancer. In response, women who had been taking hormones for years now wonder whether to pitch their pills in the trash. Whether the studies spell the death knell for the treatment remains controversial (see "Weathering the HRT Storm"); some researchers say that the trials don't match the observational results because they might not have tested groups of women most like to benefit, or that different formulations or dosages might be more effective. Studies on a potential Alzheimer's disease (AD) therapy have also produced apparently different outcomes. Numerous observational studies have suggested that individuals who take nonsteroidal anti-inflammatory drugs (NSAIDs) are less likely to develop AD. To test the idea more rigorously, researchers have conducted several clinical trials in which they gave patients who showed early signs of the disease particular doses of an NSAID, such as naproxen or rofecoxib (Vioxx). Those who took the drugs lost cognitive function as rapidly as did those who received no NSAIDs. Several factors might explain the apparent discrepancy: The trials tested particular compounds, whereas the observational studies lumped all types of anti-inflammatory drugs together, so researchers might not have picked the right one yet, says AD researcher Leon Thal of the University of California, San Diego. In addition, anti-inflammatory agents might prevent the disease in someone who's healthy but not treat it once it sets in. Further clinical studies in both cases--HRT and NSAIDs--might help researchers home in on what compounds to give and to whom.

Clinical trials progress through four stages. In phase I, researchers determine safe dosages and catalog side effects of new drugs by administering different amounts to a small number of healthy people--typically fewer than 100. If a compound doesn't have serious side effects and seems safe, researchers move to phase II trials, in which they assess whether the treatment beats a placebo or therapies currently in use in several hundred subjects with the target disease. Promising phase II results lead to phase III studies, which more accurately assess a chemical's performance by enrolling even larger numbers of subjects--usually 1000 or more. If phase III studies confirm a drug's benefits, companies apply for final approval, which certifies that the drug is effective for a particular disorder. Phase III studies aren't compulsory; in some cases, standout drugs with spectacular phase II results receive the go-ahead, especially because further research would mean denying some patients the new medication. For instance, 2 years ago, the U.S. Food and Drug Administration (FDA) approved Velcade (bortezomib) for the treatment of multiple myeloma, a bone cancer, after the resounding success of a small phase II study. But without phase III data, physicians have an incomplete guide to using the drug: The large numbers of patients in phase III trials provide a robust measure of a compound's benefit and reveal rare side effects that remain obscure in phase II. Furthermore, phase III trials often test different regimens to fine-tune the dosage. Once a medication is on the market, companies pursue phase IV follow-up studies, which probe for unforeseen side effects or additional uses for the drug, which can crank up sales.

In the United States, FDA regulates these clinical studies and approves drugs. The International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) sets standards for drug trials in the United States, Europe, and Japan to minimize redundant testing; other countries, such as Canada and Australia, don't participate in ICH but usually follow its guidelines. Foremost, these agencies protect study subjects from undue harm and determine whether study outcomes warrant a change in the way doctors treat patients.

Jumping In and Taking Off Back to Top

For every drug that clears these hurdles, hundreds--perhaps thousands--of aspirants are left by the wayside. And for every promising compound that makes it to human testing, untold numbers of eye-opening lab results fail earlier. Deciding when phenomena observed in the lab are ready for the big time is difficult, but some benchmarks are clear. Proving a treatment approach in a model animal is a crucial first step. Rodents are a common choice in the lab, but moving from even the best rodent models to humans requires a leap of faith. "Many mouse models of Alzheimer's disease don't completely replicate the disease," says behavioral neurologist Robert Friedland of Case Western Reserve University in Cleveland, Ohio. For instance, such animals might accumulate brain plaques, as human patients do, but they don't exhibit the behavioral changes--such as memory loss--that humans do. So even if a fledgling drug cleared protein goo from the brains of AD mice, researchers wouldn't know whether it also preserved memory.

Other animals can better bolster confidence that a treatment will work in people, says cardiologist Jay Edelberg of Cornell University's Weill Medical College in New York City. For instance, he's testing old pigs--whose physiology closely matches that of humans--for changes in the quantity of molecules that are connected with heart malfunctions in rodents. However, because animals such as pigs require more care and aren't as extensively studied and easily manipulated as mice are, "it's harder to find a larger animal that recapitulates human disease," he says. As a result, findings in rodents often provide the crucial impetus for clinical trials. For instance, lab studies that showed that animals given drugs called statins produce a healthier complement of cholesterol-containing particles boosted enthusiasm to test the compounds in people, and these agents are now the top-selling drug category in the United States, with sales hitting $13.5 billion last year.

View larger version (56K):
[in this window]
[in a new window]
Missing link. Although findings from rodent studies often motivate a move to humans, studies in animals such as pigs, whose physiology more closely resembles that of humans, can bolster confidence that the laboratory results will apply in the clinic. [Credit: Photodisc]

Once scientists choose a compound to test, they must enlist a different set of skills to set up an effective study. First they must clear a row of bureaucratic hurdles unique to human studies, mostly geared toward protecting trial participants. FDA calls each new compound an investigational new drug (known as an IND), and researchers must file for approval to test them in people. Next, they must prepare a protocol that describes how they will conduct the trial: how many subjects will participate, which treatments they will receive, and how long the study will last. That protocol undergoes scrutiny from an institutional review board (IRB), a panel of experts that determines whether the study falls within reasonable safety limits and whether subjects will grasp the risks involved.

Although setting up a clinical trial requires somewhat more paperwork than conducting lab studies in animals does, that part of the process is manageable, say researchers. "It's a daunting bureaucracy, but it can be handled," says Sharpless. Edelberg agrees: "From a regulatory standpoint, now instead of dealing with the animal-care-and-use committee, I'm dealing with the institutional review board," he says. The key to negotiating the system, says Sharpless, is "to make use of local expertise" by enlisting the help of clinician colleagues. Edelberg agrees, saying that clinical collaborators have helped him organize his move into human studies.

Those colleagues have also strengthened Edelberg's study design. "We work with epidemiologists to make sure that our studies are appropriately powered," meaning that they contain enough participants to reveal a given connection if it exists. "You don't want to do the study and then find out that you couldn't have gotten the answer you were looking for" because too few subjects participated, he says. Developing a good protocol is crucial because, unlike a lab experiment, you can't change it in the middle of the study, except under unusual circumstances.

Choosing what aspects of behavior, physiology, or health--known as endpoints--to assess is a pivotal decision. "The whole point of a trial is to change clinical practice," says bioethicist Jason Karlawish of the University of Pennsylvania in Philadelphia. "You want to think about measures that are meaningful to the people who take care of patients [rather than measures that interest researchers at the bench]." For instance, a promising drug might maintain muscle mass in the elderly, but a doctor is most likely to be interested in whether the compound also increases a person's mobility or prevents falls, and the most influential studies assess those types of qualities.

In addition to tracking clinical outcomes, researchers might also choose to monitor a biomarker associated with a disease, because it can save time and provide useful information. For afflictions such as heart disease, investigators want to know whether a compound reduces the chance that patients will die. But measuring the survival rate for a population takes a long time and requires large numbers of subjects. In addition, tallying only deaths doesn't reveal what's happening at early stages of disease, information that could yield insight into when a treatment is most effective. So researchers track other signs that manifest themselves sooner. For instance, artery wall thickness indicates how aggressively atherosclerosis is progressing, and changes in that characteristic can precede death by years. By choosing a readout that occurs early in disease, researchers can conduct shorter trials on fewer patients to get a preliminary view of a drug's potency before proceeding with larger studies. Similar approaches work for dementia, says Thal. Brain-imaging studies that detect hypothalamus shrinkage can provide a more precise assessment of AD progression than do cognitive tests. "To pick up a decline of one-third in cognition, you'd need about 200 subjects," says Thal. But "you need [only] 60 or 70 if you use quantitative images." Brain scans can also detect impending neurodegeneration years before AD sets in. "That [information] can give you an early indication of whether a drug was working so you can make a go or no-go decision," says Thal.

Picking People Back to Top

Once the IRB gives a protocol the green light, perhaps the most challenging part of a trial begins: recruitment. Clinical researchers must find appropriate individuals, sign them up, and keep them in the study. "It's hard for anyone who orders 40 rats from Cambridge Labs ... to truly imagine how difficult, laborious, and time-consuming human studies can be," says Murray Raskind, an AD researcher at the University of Washington, Seattle. To facilitate the process, some researchers affiliate themselves with institutions that already have patient populations under study. Clinicians also often make public recruitment pleas. For instance, the Kronos Longevity Research Institute in Phoenix, Arizona, has scrolled messages on the TV Guide channel calling for older men to volunteer for controlled testosterone replacement studies, says the institute's director, Mitch Harman. "We've gotten a terrific response," he says. But such efforts sometimes skew the populations. "People who volunteer are often atypical," says Harman. "They're more interested in life, more active. If you're looking for normal volunteers," it doesn't work. "You almost never get a random representation." Such a population could, for instance, stack the deck against an effective drug by not representing the kinds of people who would receive it after approval. Many studies of potential antidepressants give uninspiring results because of a substantial placebo effect. But "perhaps the people you really want in a study of a drug to treat depression are those people whose interest is so low and motivation is so poor that they're not going to respond or even be exposed to this kind of advertisement," says Raskind. Although the participants are clinically depressed, "you may still be missing the people for whom this drug would be exceptionally positive."

View larger version (88K):
[in this window]
[in a new window]
Sales pitch. Mass media advertisements bring patients to clinical trials, but such approaches can produce skewed test groups. [Credit: Alzheimer's Disease Cooperative Study/Porter Novelli]

Even the best recruitment strategies can't overcome the limits imposed by certain diseases. Some age-related health problems, such as stroke, don't lend themselves to planning. "With Parkinson's you can go on TV and you can have an ad and say, 'Please call our clinic,' " says stroke researcher Lawrence Brass of Yale University's School of Medicine. "You can't do that with acute stroke"; you have to go to the patients. That means conducting trials in ambulances and emergency rooms. And 40% of stroke patients have cognitive impairments, says Brass, which makes it difficult for researchers to obtain consent: a legal acknowledgement that a person understands the risks involved in participating in a particular study.

Because acuity wanes with age, getting proper consent is a challenge even in elderly people who have not just suffered a stroke. Scientists must therefore screen many more old than young patients to fill their trials. And once enrolled, old subjects can find it difficult to meet the demands, such as taking medications at the indicated times, getting to a clinic, and participating in assessments.

Other factors conspire against seniors as well. For instance, researchers design trials to produce the cleanest data possible so that if a therapy works, the numbers will show it. As a result, clinicians prefer to enroll people with only the particular disease they're trying to treat--and seniors frequently suffer multiple maladies. A cancer therapy study, for example, might exclude people with liver problems or heart disease, afflictions associated with old age. Such restrictions improve the quality of the resulting data; for instance, they reduce individual differences in drug metabolism and minimize the risk that participants will die during the trial. But because most prescription-drug users are elderly with multiple health problems, that approach doesn't provide an indication of how the treatment will work in the relevant population. So despite the fact that older people take more than their fair share of medications and many are on multiple drugs, they are underrepresented in clinical studies (see "Seniors Can Find Standard Meds Hard to Swallow"). Researchers are beginning to address the issue by devising ways to facilitate inclusion of the elderly (see "Test Patterns").

Despite the barriers that prevent older people from participating, convincing the elderly of the importance of clinical trials is easy, say some researchers. Many seniors see it as a way of getting cutting-edge treatments; others are eager to help humankind, even if they don't directly benefit, according to Karlawish.

Crunching the Numbers Back to Top

Teasing out conclusions represents the final challenge of clinical trials. Because a group of people isn't as homogenous as a group of inbred, laboratory-reared rodents, statistics--which provide a measure of confidence in the conclusions--are the cornerstone of any useful human investigation. An intervention might produce an obvious change in a population of genetically identical mice housed in the same facility and fed the same meals, but not in unrelated humans who live in different areas and have different lifestyles. "I've had people tell me that if I need [to use] statistics, it's not real," says Sharpless, because investigators used to seeing robust changes in rodents don't believe in the smaller effects commonly observed in people. And even if statistics reveal that a therapy helps patients, whether it's enough of a benefit to warrant moving forward depends on the situation. "For a devastating spinal cord injury, you might accept a modest improvement," says stem cell researcher David Scadden of Harvard Medical School in Boston--a barely detectable increase in sensation, for example. "But for Hodgkin's disease [for which successful drugs exist], you'd want a substantial gain," such as a dramatic increase in long-term survival.

Translating basic mechanisms elucidated in the lab into clinically useful therapies is an arduous task, but the endeavor is gratifying. Showing that a drug works is the ultimate proof of concept--a success that both researchers and patients can celebrate.

June 16, 2004

R. John Davenport is an associate editor of SAGE KE in Santa Cruz, California. He prefers animals to people.

Suggested ReadingBack to Top

Citation: R. J. Davenport, On Trial. Sci. Aging Knowl. Environ. 2004 (24), oa1 (2004).

Science of Aging Knowledge Environment. ISSN 1539-6150