With the stunning recall of a popular pain medication last week and the impact it had on patients, the stock market and its manufacturer, it may be worthwhile to review how drugs get to market and how safety data are gathered. Since litigation almost assuredly follows with claims of criminal corporate negligence, a perspective for the public may be useful from someone who has been a staff investigator and science project officer for the National Institutes of Health, and an investigator in approximately one hundred clinical trials spanning more than three decades.
Out of thousands of chemical compounds that may have potential therapeutic benefit, only a handful of them eventually reach the market. Before a human being ever is given a compound, extensive chemical characterization and profiling against known agents is carried out in laboratories, eventually migrating to animal studies with genetically defined mammals. Like it or not, the public should know that lowly mammals like rodents share considerable genetic similarity to humans that allows potential toxicity to be identified. With their short reproductive cycles, genetic impacts can also be uncovered. Primates, which share over 99% of our genetic code, can also be studied before a human is ever exposed to the compound. These so-call preclinical data may take 5-10 years to accumulate, and all data, if potentially favorable, are presented to the Food and Drug Administration for permission to start human studies.
Human studies are broken down into four general phases. Phase I studies are safety and toxicity studies which are rigorously monitored since no matter how similar to humans animals may be genetically, a human being's reaction may be significantly different. Many chemical compounds are dropped at this phase. If safety is confirmed, the drug moves on to Phase II studies that evaluate optimal doses and schedules of administration, e.g., once a day, once a week, etc. If efficacy and dosing are acceptable, the drug then moves on to Phase III trials where many more patients with the intended condition to be treated are identified, characterized and are included or excluded into carefully designed randomized clinical trials. Randomized clinical trials are used at essentially all phases of clinical development and will be reviewed shortly. Once all these data are acquired and analyzed to show statistical benefit, the entire development package is presented to the review body (in the U.S, the appropriate advisory committee of the Food and Drug Administration). The drug can be accepted, sent back for more study, tabled or rejected. Once approved, the drug is “launched”, but then additional Phase IV studies can be carried out to further clarify product indications, marketing positions or assess possible complications.
Since the 1970's carefully designed randomized, double-blinded clinical trials have been de rigueur to have credible data accepted by review bodies, such as the FDA. This is why almost all commercials the public sees refer to clinical trials as evidence of benefit. A population of at-risk patients is identified, without complicating conditions that may confound interpretation of results. This patient group needs to be large enough that any grouping of them has statistically the same characteristics. The new treatment is then “randomly” assigned to a group of these study patients, with the others getting randomly assigned other standard treatments and/or placebo therapy. “Blinding” the treatments means that neither the patient nor the evaluating investigator is aware of the assigned treatment. This is important to avoid any reporting or assessment bias that may occur if an individual is aware of the assigned treatment group.
Since patient safety is at the core of all ethical trials, there is a group of uninvolved scientists and statisticians who periodically review the incoming data and who are aware of treatment assignments. This Data and Safety Monitoring Committee can have sponsors of the study represented, but those individuals have no vote. It is this body that ensures the objective scientific integrity. Depending on pre-set guidelines, this review group can allow a study to continue or recommend terminating a study if unexpected benefit or harm is uncovered at an interim analysis. It is critical to know study design and makeup of the monitoring committee in order to assess the validity and significance of any clinical trial result.
If these methods were used and the FDA approved the drug, how did the data surface that caused the withdrawal of Vioxx and other FDA approved drugs in the past? Certainly there have been examples of drug companies holding back some data, but those have been rare and universally condemned.
More commonly, latent adverse effects can be missed because of the very design of clinical trials. The need to identify benefit or harm attributable to an intervention means that the study population must be statistically homogeneous. Outliers can confound information, e.g., patients with active cancer can complicate interpretation of mortality in a heart medication trial, and so cancer patients may be excluded.
Once a drug is approved and marketed, patient populations who may have not been studied in the pure populations the FDA required for approval are now prescribed the medication. These will include older or younger individuals, patients with concomitant unstudied conditions or concurrent medications or herbal compounds. Or there may be long-term effects recognized outside the short-term windows covered in clinical trials. In addition, genetic/racial differences can be uncovered in response to drugs.
When Merck's Vioxx was released in 1999, it had passed all the FDA hurdles for approval. In 2000, some early suggestions of increased cardiovascular risk emerged comparing Vioxx to another pain medication, but not in any statistically clear or valid pattern. These data were also at variance with Merck's Phase III data. Since many patients had clear benefit from Vioxx, it was important to not jump to conclusions that so-called “consumer watchdog” organizations may promulgate, until valid verification.
In 2000, Merck began a large clinical trial evaluating Vioxx for the prevention of recurrent cancerous colon polyps. During the course of follow-up of these patients, the study monitoring committee noted a statistically significant increased risk of heart attack and stroke in the Vioxx treatment group beginning at 18 months compared to placebo, though no increased death rate was observed. When these findings were reported to Merck, Merck voluntarily and immediately withdrew Vioxx from the market, and notified the FDA, healthcare providers, and the public of its action. The ethical standard Merck employed was striking and laudable, given the predictable impact this decision has on its stock valuation and its ability to potentially remain independent.
The FDA, on the other hand, will now need to reevaluate its increasingly cozy relationship with the pharmaceutical industry. It will need to resist pressure from the White House and Congress that receive significant lobbying from pharmaceutical interests. The FDA will need to demand and monitor long term studies after release. The peer-reviewed medical journals are now demanding registration of all trials so that negative trials can be reported, not just the beneficial, or positive, trials. In the future, genetic pharmacologic profiling may allow prediction of possible toxicity and drug interactions that can be tested before widespread use.
Dr. Loh is medical director of the Ventura Heart Institute in Thousand Oaks, CA. His e-mail address is email@example.com . Previous Second Opinion columns are accessible at www.venturaheart.com.