It is very difficult to select the appropriate therapy for a patient if you don’t know what disease you are treating. For the practicing physician, the patient’s presenting symptoms, history, physical examination, and radiological and biochemical evaluations typically establish the diagnosis by placing the disease in one of many accepted clinical diagnostic categories (phenotypes). The search for more clearly recognizable, homogeneous patient phenotypes has driven much of our early medical progress. Treating congestive heart failure, for example, is much more productive than trying to treat dropsy; a much older and imprecise collection of not-otherwise-specified edematous conditions.
Have we now arrived at the limit of utility of the descriptive phenotypic disease classification? I suggest that genotypic descriptions based on the root cause, or key molecular attribute, of the disease will rapidly replace phenotype-based disease classifications. This can’t happen fast enough for those in drug discovery where a drug’s mechanism of action (increasingly derived from genetic considerations) must be matched with a recognized clinical indication.
The transition from phenotype-based to genotype-based indications will not be easy. It was not that long ago that we recognized that several distinctly different genetic alterations can lead to the same clinical phenotype. For example, patients with the same clinical presentation of cystic fibrosis are not expected to respond to a therapy such ivacaftor (Kalydeco, Vertex Pharmaceuticals) unless they have the appropriate CFTR mutations among the many CFTR mutations that cause cystic fibrosis.
We have discovered that patients with the same molecular basis of disease may have distinctly different phenotypes. This means that two patients with markedly different clinical presentations may be responsive to the same therapy specifically directed at their shared molecular basis of disease. While this has yet to be reduced to routine practice, recent discoveries are clearly taking us in this direction.
For example, Kevin Strauss and colleagues at the Clinic for Special Children (Human Molecular Genetics, July 2014) have identified a variant of KCNH7 (which encodes a potentially targetable ion channel) that strongly associates with bi-polar spectrum disorder. Especially noteworthy is their observation that patients with the KCNH7 variant do not present as a single psychiatric phenotype but rather with a variety of axis 1 major affective disorders.
But medical progress in this new era depends upon coordinated activity by multiple stakeholders. In this instance, psychiatrists must be comfortable with genetic classifications of disease and be sufficiently knowledgeable to order the correct drug for patients with similar phenotypes but differing genotypes. The drug developer must have established the safety and efficacy of a new drug in patients with the specific genetic alteration and also potentially have established the lack of efficacy in patients with similar phenotypes but lacking the genotype for which the drug was developed. A diagnostic company must have developed and validated a FDA approved genetic test. Finally, there must be a reimbursement scheme that recognizes the contributions of all of the above parties.
For this to become commonplace, the clinical molecular test (at least in a prototype form) will need to exist once one begins to look for the new chemical entity that will become a drug. This will also mean that we need to invest more in genetic epidemiology. The availability of the drug for the target and the test for the target will be essential in early development, especially if there is a plan to enrich for patients with appropriate particular genotype among those with a similar phenotype.
Does this mean that all new drugs in development need a companion diagnostic? Not just yet, though we may be getting there. There is plenty of disease biology for which a drug can be made but for which a test can’t be found, including in the field of immunotherapy. But even here the secrets that regulate immune response will be revealed and genotype testing will be a prerequisite for prescription writing in this field and in almost all indications.
These and other topics will be explored at the Harvard Personalized Medicine Conference in Boston on November 12-13.
The Personalized Medicine Conference is an annual two-day event co-hosted and presented by Partners HealthCare Personalized Medicine, Harvard Business School, and Harvard Medical School in association with the American Association for Cancer Research and Personalized Medicine Coalition.
For more information and to register for the 10th Annual Personalized Medicine Conference, please visit http://www.personalizedmedicineconference.org.