Director’s Blog: Improving Diagnosis Through Precision Medicine
By Thomas Insel on
Last week the National Academy of Sciences (NAS) released a new report titled Toward Precision Medicine: Building a Knowledge Network for Biomedical Research and a New Taxonomy of Disease. This formidable title describes an ambitious effort to transform diagnosis in medicine, based on a multi-level analysis of disease, from genomic risk to the "exposome" — essentially a characterization of both external and internal exposures that can affect how a person is predisposed to disease at various life stages.
Why this new interest in transforming diagnosis in medicine? There are many examples where traditional approaches to diagnosis simply fail to define treatment response or prognosis. Moreover, our focus on manifest symptoms too often overlooks risk factors that can indicate the opportunity to prevent symptoms from emerging later.
The NAS report calls for "precision medicine," — the use of genomic, epigenomic, exposure, and other data to define individual patterns of disease, potentially leading to better individual treatment. If this sounds like personalized medicine, it is — but more so. With the over-use of 'personalized medicine' in a wide variety of contexts, "precision medicine" conveys a more accurate image of diagnosis that is person-centered and multifaceted.
The NAS report understandably focuses on cancer where diagnosis has been revolutionized by the application of molecular biology. For instance, in non–small-cell lung cancer, the traditional method of characterizing a tumor based on location and pathology has been replaced by identifying a class of genetic mutations within the tumor. These 'driver mutations' increase (or drive) the survival or reproduction of a cell, leading to the formation of tumors. For example, EGFR is a signaling protein that normally promotes cell survival and proliferation. However, a driver mutation in the EGFR gene can send the protein's activity into high gear, leading to tumor growth.
Two drugs, Gefitinib and Erlotinib, were found to have dramatic therapeutic effects by specifically blocking this abnormal EGFR activity. However, these drugs are only effective in the 10 percent of patients who have the EGFR driver mutation. These drugs appear to lack significant efficacy if tested in all patients with the traditional diagnosis of non–small-cell lung cancer. Focusing on the patient's specific molecular profile in the context of forming a diagnosis—rather than focusing only on the location and characteristics of the disease itself—is the key to choosing an effective treatment. These kinds of observations are growing in medicine, where increasing use of molecular signatures reveals that the traditional tools used for diagnosis are lumping diverse groups of diseases together.
How do we get to precision medicine? The NAS report calls for development of a knowledge management system, or information commons, similar to the geographical information system used for applications like Google Maps™. This multi-layered system would collect a broad range of health data through a vast information network (see figure) integrated into our current health care settings. Rather than considering research efforts as separate from health care, the report suggests we collect standardized molecular, exposure, and clinical data useful for research as part of routine health care. The individual patterns emerging from this multi-layered health data could define diagnosis just as multi-layered geographical data can define position. The long-term vision builds on information technology to provide new maps that guide patients and clinicians.
The NAS report focuses heavily on cancer, but the implications for research and diagnosis of mental illness are important. Depression, schizophrenia, borderline personality disorder, and autism spectrum disorder are complex syndromes. It may be that many different disorders are embedded within each of these categories. The lesson from other areas of medicine is that a diagnosis that relies solely on manifest symptoms is not the best guide to choose the most effective treatment.
Precision medicine for mental disorders could be even more transformative than for cancer. Will subdividing syndromes based on molecular signatures, neuroimaging patterns, inflammatory biomarkers, cognitive style, or history give us subgroups that are more responsive to certain medications or psychosocial treatments?
It's a question worth answering. The current debate about the efficacy of antidepressants might be settled by defining the subgroup that responds. Recall the lesson of Gefitinib and Erlotinib—just because a medication is not effective for everyone does not mean it will not be effective for anyone. The task is to identify the biomarker that predicts response—whether the treatment is a medication or a psychosocial intervention.
The Research Domain Criteria (RDoC) project is one important step toward precision medicine for mental disorders. By building a classification system based on many layers of data and adding dimensional approaches rather than categorical labels, RDoC should begin to give us the precision currently lacking with "symptom-only" diagnosis of mental disorders. How the RDoC scheme will align with treatment response remains to be seen. But as the new NAS report points out, the process begins with collecting the data. What we are learning from cancer is that this approach brings not only precision but hope.