Incidental research imaging findings: Pandora's costly box.

A healthy adult volunteers to serve as a control subject in an fMRI study. The subject provides informed consent, and a tailored protocol designed for the particular study then is executed. In most cases, the research subject altruistically assumes that the result will benefit science in some way. However, does science owe the volunteer a fair assessment of the information not relevant to the focus of the investigation (i.e., should the images be read by a capable individual who has facility to detect abnormalities and can render an opinion of what constitutes normal)? In the article by Illes et al.1 in this issue of Neurology, the incidence of abnormal findings in 151 research scans, requiring physician referral, was 6.6%. The authors state that most of the findings were not detected or reported to the participants. This is a problem and highlights several issues associated with imaging research. First, some MR research—particularly fMRI—is performed by scientists who are not skilled at image interpretation. Thus, most “incidental findings” will not be detected under the designed protocol. However, to miss a large asymptomatic posterior communicating aneurysm and have the patient return a year later after a subarachnoid hemorrhage asking, “Was my aneurysm not visualized on the previous study?” is an ethical dilemma. What is the added cost of having an imager view and report the findings? There is clearly a cost in dollars and time. Nevertheless, it seems small compared with the price one pays for not detecting significant lesions. Additionally, some lesions, such as certain white matter abnormalities, may actually confound the results of the scientific study. Should not the investigator want the images read by a credible expert? If these are “healthy” subjects, then imaging data should be interpreted to confirm this stipulation. When an important abnormality is present, the subject trusts that the research team will observe it. Ethically, we owe such consideration to our volunteers, although this is not acknowledged at the signing of the informed consent. A second issue is the protocol itself. What is standard for a researcher may not be sufficient for an imager. Time and cost probably prohibit comprehensive protocols. This limitation should be stated in the informed consent. The reader should interpret the study as is. This is a generally accepted principle in radiology. To do otherwise would be prohibitive to efficient study design and a major impediment to research. Third, if an abnormality is detected, who should receive the results, and in what setting should they be discussed? This is complex question regarding confidentiality, the patient, and the caregiver. However, there needs to be a protocol endorsed by the institutional review board that addresses these circumstances. Fourth, investigators must be educated concerning incidental findings. Some basic science researchers may not be aware of the scope of abnormalities that may be detected in “asymptomatic” volunteers by individuals with expertise in imaging. This fact impacts the subject and the research data. Finally, there are established rules of ethics that require disclosure as an element of consent to enter the study. The ethical requirements of informed consent for volunteering to serve as a control subject include a clear understanding of the risks of participation. In a brain fMRI study, the risks are not simply the trivial risks of undergoing an fMRI. More importantly, they include 1) increased anxiety at being told there is or may be an abnormality discovered; 2) financial costs associated with additional diagnostic testing to pursue an apparent abnormality; and 3) health risks resulting from the medical complications of additional testing done to clarify the abnormality. These points should be made clear in the written consent form as potential risks of participation in the study. Otherwise, the consent is invalid.