Incidental findings on brain magnetic resonance imaging

Incidental findings on brain imaging are defined as previously undetected abnormalities of potential clinical relevance that are unexpectedly discovered and unrelated to the purpose of the imaging. Incidental findings are increasingly detected in clinical practice, with screening, and in the research setting. Data on the prevalence of these abnormalities are scarce, the clinical course of the findings is often unknown, and—as a result—the management of such lesions is not clear. As a community, we need to set standards on how to deal with such findings.

In the linked study (doi:10.1136/bmj.b3016), Morris and colleagues systematically reviewed the literature on incidental findings on brain magnetic resonance imaging (MRI) and found a prevalence of 0.7% (95% confidence interval 0.47% to 0.98%) for neoplastic lesions including meningeoma (0.3%), and 2.0% (1.1% to 3.1%) for non-neoplastic lesions including aneurysms (0.4%).1 The authors state that they have increased the precision of existing estimates on the prevalence of incidental findings.

However, the findings have merely touched the tip of the iceberg. The 16 included studies were published between 1989 and 2008, and the incidental findings were detected on scanners with different field strengths and using different imaging protocols. The rate of detection of incidental findings depends on the image sequences in the MRI protocol (including the use of contrast), the experience of the readers, the use of a prespecified analysis protocol, and postprocessing of the images. For example, without bright blood or black blood imaging sequences, aneurysms can hardly be detected. Protocols without contrast agents will miss small lesions, such as meningeomas or vestibular schwannomas. Interpreting images with a checklist of predefined lesions will lead to higher detection rates for those lesions. Multiplanar reconstruction of images can enhance the detection of some pathologies—for example, coronal projections enhance the detection of hypophyseal macroadenomas.

In clinical practice, the MRI protocol is aimed at solving a specific clinical problem. In a research setting, the MRI protocol is dictated by the scientific problem and the research questions. Thus, the prevalence of incidental findings can be expected to vary depending on the purpose of the MRI.

The pooling of data from different settings and purposes will therefore not lead to a better estimate of the prevalence of incidental findings. This is especially true when older studies using imaging sequences with low resolution and scanners of low field strength are included. With an optimal MRI protocol the prevalence of a specific pathology will approach the “true” prevalence based on data from autopsies. Indeed, a recent high resolution imaging study found a prevalence of aneurysms approaching that reported in autopsy studies (1.8% v 2.0%),2 3 which is substantially higher than the prevalence figure presented by Morris and colleagues.1

With improvements in imaging technology (higher field scanners, new pulse sequences), the number of detected incidental findings will increase dramatically. Another important point is the advent of imaging biomarkers. Advanced techniques for postprocessing and analysis, such as automated segmentation of brain structures or voxel based morphometry, will lead to the discovery of imaging biomarkers—for example, in dementia or Parkinson’s disease. Once the predictive value of these markers has been established, most MRI studies of the brain, both in the clinical and research setting, will reveal information that might be relevant for the wellbeing of patients or participants. Although still incidental, these findings can unfortunately no longer be considered unexpected. We will soon face large medical, ethical, and practical problems as a result of technical improvements.

In general, management of incidental findings in the clinical setting is not a big problem. The MRI images are reviewed by experienced radiologists, the patients have a relationship with the referring doctor, the findings are usually less severe than the clinical problem for which the patients were referred, and follow-up of the patients can easily be implemented.

Incidental findings in the research setting are a much larger problem, and international guidelines are desperately needed. The debate in the United States is far ahead of that in Europe and other continents.4 5 6 In a workshop sponsored by the National Institutes of Health and Stanford University in 2005, and attended by scientists, doctors, lawyers, and ethicists, the facts and problems related to this topic were explored.4 It became clear that investigators vary greatly in the way they handle incidental findings.7

Several important questions remain. Do researchers have the obligation to examine each brain scan for the presence of incidental findings? Should MRI scanning always be conducted in accordance with standard medical practice, even if this is not necessary for the research question? Should all scans be read by an expert in neuroradiology? If an incidental finding occurs what should researchers do, and how much and when should the participant be told? What should we do if participants were not told about incidental findings, such as microbleeds or white matter lesions, because the clinical course was not known at the time of the study if it later becomes apparent that the findings greatly increase the risk for disease? How should we deal with newly discovered imaging biomarkers—are we obliged to examine previously acquired scans for the presence of these biomarkers? What is the role of institutional review boards and funding agencies in setting regulations?

Since 2005 several practical approaches to incidental findings in the research setting have been published.6 8 The most important advice is that studies that involve brain imaging should detail exactly how incidental findings will be handled, and this information should be provided to the institutional review board and explained during the informed consent process.

Overall, the line between the large responsibility that doctors have for their patients and the minimal responsibility that researchers have for participants is being challenged. It is our responsibility as clinicians and researchers to reach a consensus on the optimal protocol for the management of incidental findings.