We need a massive expansion in phenotyping

A key part of genetic medicine involves investigating if genetic changes (genotypes) have caused measurable human changes (phenotypes). To do this we need information on both genotypes and phenotypes. We have seen a massive expansion in genotyping in the last five years, and we can now quickly and easily generate millions of genotypes from people. But we do most phenotyping in the same way as ten years ago. We urgently need a massive expansion in phenotyping to match the massive expansion in genotyping that DNA sequencing technology has delivered.


What is phenotyping?

Phenotypes are the physical or biochemical human traits that result from genetic or environmental influences, for example our height, eye colour, or diseases. Phenotyping is the recording and measurement of these traits. Phenotyping covers a much broader set of activities than genotyping. Assessing intellectual development, measuring height and weight, recording someone’s haemoglobin level, or taking an X-ray to evaluate heart size are all examples of phenotyping.


Phenotyping is much harder than genotyping

If you are born with a genetic variant you will have it for the rest of your life. This means a one-off whole genome sequencing test will allow genotyping of all 3 billion DNA bases that make up your genetic blueprint. But your phenotypes can change a lot during your lifetime; height increases rapidly in childhood, diseases can occur at any time. We can also intervene to change phenotypes in a way that is not possible for genotypes. For example, someone with a BRCA1 mutation might have developed ovarian cancer, if she had not had her ovaries removed. These complexities mean a one-off phenotyping exercise often has limited value.  And we often need extra contextual information to use phenotype information appropriately.


How is phenotyping done?

Phenotyping is done differently for different traits. But the standard way to phenotype has remained the same for many years: An individual health professional acquires the relevant phenotype information from an individual patient by examination, measurement or questionnaire. Some impressive examples of large-scale phenotyping do exist such as UK Biobank, which collected baseline phenotyping measurements from 500,000 individuals and has recently completed genotyping in these same individuals. DECIPHER is another wonderful example and includes phenotype information from 18,000 individuals with rare disease from 250 centres. But these examples are very much the exception and we need to make systematic phenotyping routine.

How can we achieve this massive expansion of phenotyping?


Harnessing technology

Too much phenotyping is done manually. It’s laborious, costly and error-prone.

We can and should make more use of technology for phenotyping. This is starting to happen in some areas. For example, digital images are being increasingly used to collect and analyse facial phenotypes, which are often distinctive in genetic diseases.

We also need to develop more automated processes for storing and sharing phenotypes. Far too much phenotyping relies on manual entry of information to databases, often multiple times to different databases. This is very laborious, very costly and error prone.

Passive phenotyping, whereby information is collected without any human input, will probably be central to the phenotyping transformation we urgently need. There are some exciting innovations in this area. Mindstrong Health is one example. They are exploring using the passive, continuous interactions of humans with smartphones to develop quantitative measures of brain function in health and disease.


Harnessing global connectivity

Many genetic diseases are very, very rare. Previously this made collecting phenotypes very challenging. But in the era of Facebook and Google, patients can find each other, however rare the disease. There are Facebook community pages for hundreds of rare genetic diseases, and various websites act as a forum to link people with rare diseases, such as, NORD, RareConnect and Genetic Alliance. As well as providing valuable support, these forums mean we can now find and get phenotype information from more patients with genetic disease.


Harnessing patient power

Harnessing the expertise of patients is probably the most important thing we need to do. Phenotyping has traditionally been a somewhat paternalistic enterprise. A health professional would acquire the information from a patient and enter it into a database. Wherever possible we should now remove the middleman (usually middlewoman) and develop systems that let the patient directly submit their information. Many patients greatly welcome this opportunity and patients are undoubtedly the experts in how a disease affects them!


Harnessing patient reported outcomes

Genetic medicine still mainly uses phenotype information to aid diagnosis. We try to match genotypes and phenotypes to work out the cause of someone’s illness. But in the near future the focus will start to shift to using phenotypes to aid treatment, such as developing new drugs. For this to be successful we need to have objective measures to judge whether a treatment has worked. For some conditions, like cancer, this is fairly straightforward – if the cancer shrinks the treatment has had an impact. But for many conditions it is much harder to decide how to measure effectiveness of an intervention. Including patients in all such evaluations is essential.  We need patients to tell us what problems their genetic condition has caused and to help us devise objective measures to gauge if a treatment has had an impact on these problems.

Cystic fibrosis is an example of how this can work successfully. Improvement in patient reported outcomes were central to the approval of recent new therapies for cystic fibrosis. These new drugs can be used in treat nearly half of patients with the diease.


Concentration, communication, integration,

To achieve this phenotype revolution we must concentrate some time and energy on defining our key phenotyping objectives. Then we must foster communication between the groups needed to deliver the objectives. Because the complementary knowledge of patients, doctors, scientists and many others  will all be needed for us to be successful.