safety team’s role is to quantify and set priorities among potential risks posed by the drug they are reviewing. They do not make recommendations on a drug’s approvability (Racoosin, 2006). The committee was told that discussions have occurred in FDA about including a full-time safety officer in the other ODS/ODE divisions. Two possible explanations for why that has not occurred were offered: shortage of safety officers and the fact that some divisions do not review enough applications to support a full-time safety officer.

Completion of Clinical Trials and Their Limitations

Clinical trials typically take 2–10 years to complete (PhRMA, 2006), depending on such factors as the rate of the event of primary interest, the length of patient followup, the staging of trials, and the difficulty of accruing patients. When data from phase 2 trials seem extremely promising, particularly in the context of serious or life-threatening diseases or conditions, an NDA may be filed without proceeding to or completing phase 3 trials.4 For example, azidothymidine (AZT) was approved for treatment of HIV infection on the basis of phase 2 trials (Grassley et al., 2004).

Only about 20 percent of drugs that enter phase 1 trials go on to be approved and marketed (lower estimates have been provided by others so the true percent is unknown) (DiMasi et al., 2003; PhRMA, 2006). A sponsor may decide to halt trials for various scientific or commercial reasons. A recent study found that the number of clinical trials being conducted in the United States leveled off in 2000 and then started to decline in 2002. The author attributes the decline to cancellation of late-stage trials (Kaitin, 2005).

Even when a sponsor completes its trials and submits the resulting data in support of an NDA, important safety information about the drug is not yet available. That point is essential for an understanding of the drug regulatory system and the incomplete safety profiles of the drugs that enter the marketplace.

The gaps in critical information, such as safety data, are due to a number of factors, including the limited number of subjects studied and the ways in which the subjects and the research setting differ from the conditions of use when the drug is marketed (CDER et al., 1998). Preapproval trials typically are too small to detect even significant safety problems if they are rare. An adverse event (even a serious one) that occurs in less than one in 1,000 patients cannot be reliably detected except in the largest premarket trials but can pose a serious public health problem when hundreds of thousands


FDA may grant accelerated approval on the basis of surrogate endpoints, but approval is conditional on sponsors’ undertaking or completing validation trials.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement