ABSTRACT
A new treatment in a patient population may decrease or escalate the risk of one or more serious adverse events (SAEs) over time. Escalation of such a risk in particular is very important to detect as quickly as possible for the patients’ safety sake. Risk may remain constant over a relatively short period of time but then suddenly it may start to take a wrong turn, that is, start to escalate over time. On the other hand, a new effective treatment may completely stop or decrease the progression of a serious disease, such as cancer, over time. Such a drug can be a lifesaver for many patients. In drug development it is important to learn about such treatment-related benefits and harm that are time dependent. This is usually best understood through randomized controlled clinical trials (RCTs). RCTs invariably collect efficacy and safety information on treatments under study through the so-called time-to-event endpoints. A time-toevent endpoint is a measurement of time for an event from the start of the treatment until the time that event occurs while the patient is in the clinical trial. Some examples
of time-to-event endpoints on this topic are summarized in Table 8.1 and discussed in detail below. An example of a safety time-to-event endpoint is the thrombotic cardiovascular
event endpoint which can be the first occurrence of any one of the events of myocardial infarction, unstable angina, cardiac thrombus, resuscitated cardiac arrest, sudden or unexplained death, ischemic stroke, or transient ischemic attacks. The vioxx gastrointestinal outcomes research (VIGOR) trial [1], a clinical trial evaluating VioxxTM (rofecoxib) in arthritis patients, raised the issue that the use of this drug may escalate the risk of such a safety event over time. In this case, the safety event is a composite of several related cardiovascular events. The adenomatous polyp prevention on vioxx (APPROVe) trial [2], a second clinical trial evaluating rofecoxib in colorectal polyp prevention confirmed that the risk of cardiovascular events escalates over time. Another example of a safety time-to-event endpoint is invasive breast cancer, found in a landmark clinical study by the Women’s Health Initiative designed to clarify the risks and benefits of combination hormone replacement therapy of estrogen plus progestin [3]. This study was originally designed to evaluate the treatment benefits of prevention of coronary heart disease over a period of 8.5 years. However, the study was stopped prematurely inMay 2002 after mean follow-up of 5.2 years because the study data showed that postmenopausal women receiving combined estrogen plus progestin had an increased risk of invasive breast cancer compared with placebo. Some argue that the critical points where this risk increases is around 4 years after treatment and again around 6-7 years after treatment [4]. In many clinical trials, safety events are not observable directly for several reasons;
the length of follow-up in a trial is relatively short, the sample size too small, or the event is rare enough to be observed in the planned sample of the clinical trial. Often RCTs make conclusions about the safety of new treatments based on clinical judgments and by indirect means given the limited number of exposed patients in clinical trials. Signals of a potentially serious adverse event (SAE) may be detected through
biochemical markers such as abnormal elevations in laboratory tests. An example in this regard is the safety event of renal toxicity which can be detected through
TABLE 8.1: Examples of time-to-event endpoints with changes in hazard.
