Outcome Measures or Endpoints for Clinical Trials
The appropriate selection of outcome measures for cancer trials is a focus of ongoing debate. While recognizing that proper outcome selection is influenced by the specific design characteristics and resource limitations of the study, the National Breast Cancer Coalition believes that endpoints should be chosen in a manner that maximizes the usefulness and meaning of the clinical trial in terms of addressing the burden of breast cancer for women.
The natural desire for quick answers and less costly studies is typically at odds with the desire for ideally valid and reliable measures of effectiveness and safety. While the selection of less definitive endpoints may allow for shorter trials and easier measurement, they may not yield useful information about how well an intervention saves or prolongs lives. Whenever less-than-ideal endpoints are selected, important questions will be left unanswered. This means that further trials will be necessary – requiring more time, more money, and more patient participation. Such a waste of resources can be avoided by selecting, at the outset of testing, outcomes that provide the clearest indications of the true efficacy and harms of the intervention being examined.
The National Breast Cancer Coalition believes that trained consumer advocates must be involved in the design and monitoring of clinical trials. Women who have a personal experience with the disease and have educated themselves to understand the scientific and clinical research concepts of breast cancer bring an informed and critical perspective that makes clinical research accountable to the patient community and to funders of research, by ensuring that trials ask truly important questions, and that measured endpoints reflect the true benefits and safety profile of interventions under investigation.
The purpose of breast cancer clinical trials is to measure the safety and effectiveness of specific interventions used against the disease. Whether considering interventions related to prevention (stopping cancer from occurring), screening and detection (identifying cancer when it occurs), or treatment (examining the potential positive and negative effects of drugs, devices, procedures, or other approaches), well-designed, randomized clinical trials are critical to improving breast cancer care. Regulatory agencies, clinicians, and patients rely on clinical trials because they produce the most reliable evidence of the true value of possible treatments or interventions.
Before a clinical trial begins, researchers must determine what outcomes or endpoints they will monitor in order to be able to reach meaningful conclusions about the effect of the intervention they are studying. Because survival is the most important issue for most patients, mortality is the major or primary outcome of interest. From this perspective, the best test for any intervention would follow study participants to compare the mortality rates between those exposed to the intervention to those not exposed at pre-specified time points into the future. Depending on the type of cancer and the effectiveness of the intervention, a trial like this can be quite expensive in terms of time and resources, and as a consequence, other methodologies and other endpoints are sometimes chosen.
In order to shorten the time needed to reach conclusions about an intervention under study, and to collect additional information that might be used to help interpret the primary outcome, many cancer clinical trials also measure interim or secondary outcomes such as disease-free survival or tumor response rate. Sometimes the secondary outcomes allow researchers to see trends in an intervention’s effect earlier in the trial. The researchers may even agree in advance that if the interim outcomes reach pre-specified levels, the trial will be concluded early, the positive or negative interim outcomes will be reported, and conclusions about the ultimate effect of the intervention on mortality will be determined.
Measurement of interim outcomes can result in an increase in the number of “events” observed and a decrease in both the number of patients needed in a study and the length of time those patients need to be followed. It is important to understand, however, that the validity of this approach relies on the assumption that a favorable interim outcome actually correlates with lower mortality. Unfortunately, sometimes an intervention that has a beneficial effect on the interim outcome might ultimately be proven to have no effect or a negative effect on survival, or vice-versa. If the study were ended without collecting the mortality or survival data, the overall effectiveness of the intervention would be misunderstood. The importance of an error of this type can best be assessed in the context of the possible positive or negative impact the intervention has on the quality of life for the patient, and the resources that are wasted when a trial produces unreliable results.
Sample Size and Selection of Outcome Measures
The size of the available patient sample and the need to avoid experimental bias also play roles in determining which outcomes or endpoints will be chosen for a given clinical trial design.
Finding the right sample size ensures that time and resources won’t be wasted in testing too many or too few patients. Estimates of how many participants will be needed to ensure that the trial results are meaningful are based, in part, on what primary and secondary or interim outcomes are going to be measured. This means that when patient availability is limited, the choice of endpoints might also be limited. Generally speaking, the more likely the primary outcome and the greater the expected difference in outcomes between the patients getting the intervention and those who don’t, the smaller the number of participants who will be needed. At the same time, if the sample size is too small, there is no way to be confident that the intervention would produce the same outcomes in a larger, more typical population.
Discovering the true effect of an intervention also depends on ensuring that the trial is not biased by unrecognized factors that influence the study outcomes. Increasing the sample size is one way to guard against some sources of bias, and this, in turn, might influence outcome or endpoint selection.
Measuring Survival in Breast Cancer Clinical Trials
Measures of mortality and survival track the success of an intervention in preventing cancer deaths. The mortality rate is expressed as the number of deaths in a certain period of time per standard unit of population. Survival usually refers to the number of people alive for a given period after an intervention.
Mortality or survival endpoints might be based only on deaths resulting from breast cancer (cause-specific mortality) or on all deaths in the study population regardless of cause (all-cause or overall mortality). All-cause mortality can be a particularly useful outcome measure where interventions might have both beneficial and harmful effects. For example, the intervention may decrease breast cancer mortality but increase endometrial cancer mortality. As long as the cause of death is accurately assessed, both of these results would be discovered. Mis-classification error occurs when the cause of death among trial participants is incorrectly identified, leading to inaccurate conclusions about the effect of the intervention on mortality or survival.
Because mortality and survival are critical, easily measured, and objective endpoints, and because they can allow researchers to detect both beneficial and harmful effects of an intervention (decreases and increases in deaths), they are usually considered the gold standard for measuring efficacy in clinical trials of cancer treatments. In some studies, however, their usefulness as single endpoints would be limited. For example, mortality would not be a useful endpoint in trials where few deaths are expected, or where the interventions under investigation are expected to have small or moderate effects. Similarly, for cancers where most patients live well beyond the time of diagnosis and/or trial participation, it can take many years to gather sufficient data on mortality or survival to accurately assess an intervention. In such cases, interim endpoints can be important, but survival data must still be collected in order to fully assess the ultimate usefulness of the intervention. This should also be expected when trends in interim outcomes lead to a decision to end a trial early.
Other Endpoints Common to Breast Cancer Trials
Incidence – Defined as the number of new cases in a population over a defined period of time, the incidence is typically chosen as an endpoint in prevention or screening trials. For example, the effects of exercise, environmental exposures, medicines, food supplements, behavior change, preventive surgery, education, or a combination of these, might be compared with one another, and with no treatment, placebo, and/or a standard intervention to lower the incidence of breast cancer. Incidence can also provide valuable information about unintended effects of an intervention. This might occur if a treatment designed to prevent an unrelated medical condition is found to increase breast cancer incidence.
Time to progression (TTP) – TTP is the time from the start of an intervention until an event such as an increase in tumor size or a metastasis is detected. TTP is typically measured in treatment trials involving advanced breast cancer or metastatic disease. Trials using these endpoints usually require a smaller number of patients and shorter patient follow-up than those measuring mortality or survival, but they do not provide information on possible longer-term effects of the intervention.
Tumor response rate – Referring to a measured reduction in tumor size as determined by x-ray or other radiographic imaging, tumor response rates are often used as an outcome measure in clinical trials involving metastatic disease. Tumor response rate is also used in neoadjuvant trials, to assess the effectiveness of treatment in shrinking a tumor before surgery.
Recurrence – cancer can be local or regional (in the area around the breast), or distant (in other areas of the body). Recurrence is a reappearance of breast cancer during or after treatment. Recurrence is expressed in terms of the number or proportion of participants who have had a recurrence at a selected follow-up time point.
Disease-free survival (DFS) – Another way to look at recurrence is disease-free survival which is also referred to as event-free survival. DFS is the time from the beginning of an intervention until a patient experiences a recurrence, a new primary cancer or death.
Progression-free survival (PFS) – Progression-free survival is the time from the beginning of an intervention until a patient shows signs of disease progression.
Women need to be able to make health care decisions based on the highest possible quality of evidence. To achieve this, we need well-designed clinical trials that use the most valid and meaningful outcome measures available. We recognize that selection of the ideal experimental endpoint is not always possible or practical. On the other hand, designing a trial with practical but inadequate outcome measures will only result in wasted resources and delays in identifying the most effective healthcare. Before a clinical trial begins, researchers will decide what outcomes or endpoints they want to measure.