McDonald and Sadanand: Clinical trials, point-counterpoint

As difficult as it may be to deny a drug to a desperate patient, controlled trials are the backbone of clinical science: We need them to answer the basic question of whether a drug works, and for which patients. Trials are not primarily designed to help the patients who participate in them, nor should they be. The fact that some physicians have seriously argued for switching out control patients who are not responding to traditional treatments — essentially destroying the trial — is disconcerting. Instead, trials should be designed to give the best possible estimate of the overall benefit-to-risk ratio of a drug, without which clinicians cannot make informed decisions about prescribing the medication to future patients.

Randomized controlled trials seem to become an ethical issue only when they involve terminal patients vying for especially promising drugs. The visceral pang of wrongdoing is only there, however, because one has assumed the drug is effective, and the stakes are so high. Unfortunately, less than half of the drugs tested at the Phase III level, typically the last stage before FDA approval, outperform a standard treatment. One prominent example of a drug that failed its first Phase III trial was the laboratory wunderkind Avastin, which prevents the growth of blood vessels to feed tumors. The drug made intuitive sense to scientists, who knew cancer growth was far more difficult without a direct food source, but it did not improve outcomes for breast cancer patients in its first Phase III trial. The high failure rate of drugs even at late-stage trials means that on average a trial patient will actually live longer when participating as a control subject.

The cold calculus of debating which patient groups — the control group or the experimental one — are the ‘lucky’ ones misses the point that only the trial itself can tell us this information. Altering study design, such as using previous control data, risks producing inaccurate or inconclusive data that leads to unsubstantiated conclusions.

It certainly would be unethical to provide treatments for years to patients that do not actually work, or work less well than we think. Doing so would delude patients when they make critical decisions about their health, misappropriate tax dollars and misguide research. It is this latter point that concerns me the most; if we think we’ve found the magic bullet, we will not put any more effort into research.

It is understandable that oncologists want to use PLX4032 for palliative care, since it seems to ease the symptoms of late-stage cancer, at least anecdotally. I am not against using the drug in this capacity but I do object if the drug is never fully tested to determine whether it increases lifespan. It is too easy, even for physicians, to be swayed into thinking a drug must extend life if it decreases tumor size and provides symptomatic relief.

As more drugs are tailored to specific diagnoses, such as certain mutations in cancers, it is increasingly likely that trial drugs will be more effective for selected patient groups. We shouldn’t view this as a problem, but the success of modern biomedical research is that the system remains inherently utilitarian. If the likelihood that the experimental drug will work is high, at most several hundred patients may die, while thousands or more new patients will live because of the former’s sacrifice.

While the Food & Drug Administration and drug companies can work to move trials along more quickly, the most prudent way to minimize loss is to perform the absolutely best science possible — and that means not skimping on control groups.

Jessica McDonald is a fifth-year student in immunobiology.

///

On a human level, everyone — doctors, the patients and their families, the Food & Drug Administration — can feel for a family who loses a loved one in a drug trial, but sadly the story of Brandon Ryan is merely representative of the agony many patients’ families are forced to unfairly endure, highlighting that clinical trials such as these are fundamentally flawed and must be changed.

We should not eliminate controlled studies. They are vital to ensure that patients are receiving the best possible treatments. Equally importantly, they provide a safeguard — albeit a not always effective one — against unethical drug company practices by allowing the FDA to evaluate a complete body of data before determining whether a drug should be allowed on the market. However, it is clear that controlled studies need to be re-optimized to ensure both efficacy and maximum patient comfort.

The PLX4032 drug trial was problematic in two ways. First, patients placed in the control group were aware that they were not receiving PLX4032, as were the doctors assigned to follow their progress. The study organizers justified this decision because PLX4032 is delivered orally as opposed to standard chemotherapy infusions, making it easy for patients to figure out whether they were receiving the new drug. Because an adequate placebo treatment could not be devised, the study was not sufficiently controlled, rendering the treatment given to the control group pointless and cruel. The scientists should have instead used clinical data from past cases in which patients only received the older chemotherapy regimen.

A second problem with this study was that it was apparent to both participants and doctors that PLX4032 was already far more effective in earlier trial stages for reducing tumor size than the older chemotherapy treatment. Given this, the study organizers should have instituted a shorter evaluation time for this study. In fact, drug safety monitoring boards — which evaluate trials such as these — should have stopped this trial, given there was no longer any uncertainty about the safety and benefits of PLX 4032 over the traditional chemotherapy treatment.

In the last decade we have come to realize how complex cancers are. While there are some gene mutations that substantially increase the risk for cancer most cancers cannot be linked to one gene mutation but rather to a slew of mutations. This complexity means that cancer drugs may not be useful for a broad range of patients, but many only work effectively for a subsection of patients. Doctors and the FDA need to recognize that this it is unrealistic to expect cancer drugs to treat everyone. It is also impractical to use longitudinal studies such as these, where lifespan is the only major measurement, to evaluate cancer drugs. It is time for the medical community to reevaluate the set-up and evaluation of drug trials such as that for PLX4032 so that more families do not have to suffer due to nothing else but the bad luck of being assigned to a control group.

Saheli Sadanand is a fourth-year student in immunobiology.

Comments