What's in a phenotype? Next-generation genomic diagnosis
7 September 2015
Phenotyping: a tale of two settings
The advent of next generation sequencing technologies has resulted in an explosion of genomic data and discovery of novel genetic mutations (variants). The challenge for genomic investigation is no longer the generation of sequence data, but its interpretation. Genomic data in isolation from contextual clinical and phenotypic data are insufficient for unravelling the complexity of genetic variation and understanding its biological relevance. As a result there has been a flurry of initiatives to collate human phenotypic data on a massive scale, including the Google Baseline Study, the 100K Wellness Project, and Human Longevity Inc.
However, in contrast to most research-focused endeavours, phenotyping within the context of a clinical diagnostic service must contend with the constraints of a frontline healthcare setting, from finite resources, limitations on clinician time and often diagnostic urgency. Our latest briefing in the Clinical Genome Analysis series outlines the importance of phenotyping for genetic and genomic rare disease diagnostics, and highlights the urgent actions needed to ensure the phenotyping process in a clinical setting is effective and robust enough to support genomic diagnostics and benefit patients. Key actions include the considered integration of standards to describe phenotypes, tools to enable systematic capture, improved exchange of phenotype data and a pragmatic assessment of what data to collect under different clinical circumstances and at what point along the diagnostic pathway.
What is phenotype data?
Phenotype data can be both quantitative (body weight, height), and qualitative (behaviour) and in the context of rare disease diagnosis, clinicians may require phenotypic information on the patients relatives as well as the patients themselves. Phenotypes can encompass photos, descriptions of clinical features, medical histories, physical or biochemical measurements and observations. The absence as well as presence of features can be equally important in collating a comprehensive clinical picture to aid diagnostic interpretation of variant data. Tests to gather phenotype information can be invasive (e.g. blood tests, or tissue sampling), as well as non-invasive (e.g. measuring the heart rate or undertaking an X-ray). Decisions as to whether and when to undertake some of these examinations must therefore be balanced with the relative clinical value in each patient case and circumstance.
Phenotyping for clinical diagnostics: how much, how detailed and when?
Phenotypic information is essential at several stages of the diagnostics pathway, both upstream and downstream of genetic testing. Upstream of the test the information can help focus the disease area under investigation, inform choice of subsequent clinical tests and, if clinically indicated, inform the choice of an appropriate genetic test. Downstream of genetic testing, phenotype data along with family history is necessary to filter, prioritise, and interpret potential disease causing genetic variants.
Our briefing illustrates the different approaches to phenotyping; ‘high-level’, ‘broad’ or ‘deep’. The most appropriate and effective phenotyping approach depends on the circumstances of each clinical case. Sometimes ‘high-level’ phenotyping is sufficient to determine the choice of genetic test and data interpretation. In other cases, the different approaches to phenotyping may be warranted at different stages of the diagnostic pathway depending on the outcome of the genetic test.
Take for example a patient presenting with an aortopathy (disease of the aorta). Multiple genes a re involved in causing aortopathies and there are several conditions associated with overlapping clinical manifestations. So a careful search has to be made to ascertain broad phenotypes and ensure these are captured to help interpret results as well as undertaking the correct genetic test. To begin to focus the potential genetic cause of the aortopathy, the clinician may consider whether the patient has Marfan’s syndrome. ‘High-level' phenotyping, involving non-invasive, physical examinations for skeletal features as well as of eye and heart features can determine whether the typical features associated with Marfan’s are present in the patient. If the phenotype assessment using standard and accepted clinical diagnostic criteria along with family and medical history is indicative of Marfan’s, then the clinician would proceed to order a genetic test for the fibrillin (FBN1) gene to confirm the suspected diagnosis. Mutations in this gene affect the function of the protein fibrillin-1, essential for structural support in connective tissue throughout the body. If a pathogenic mutation is not found in FBN1, it may be necessary to perform further phenotyping, this time broader and deeper, covering the spectrum of clinical manifestations that might accompany aortopathies in order to further focus the choice of genetic test and subsequent data analysis and interpretation.
Loeys-Dietz syndrome has features similar to Marfan’s syndrome, but is caused by mutations in the TGF-beta receptor genes (TGFBR1, TGFBR2) rather than FBN1. Conversely mutations in the FBN1 gene can cause conditions other than Marfan’s syndrome, for example Ectopia Lentis syndrome which shares some, but not all, features with Marfan’s syndrome - even if mutations are found within the FBN1 gene it can be appropriate to cross-check phenotypes with associated syndromes.
Realising genomic medicine
In healthcare settings, phenotyping and genotyping do not strictly follow a unidirectional course and it is not always possible to anticipate the types of clinical tests necessary to solve a diagnostic puzzle before performing the genetic test itself. In fact, collecting comprehensive and detailed phenotypic information prior to the genetic may not always be attainable or essential in some cases, and indeed doing so when unnecessary may have negative consequences. These include deflecting limited resources from other clinical needs, delays in patients being assessed and delays in patients being referred for genetic testing.
Undoubtedly, the full potential of genomic data can only be realised by integrating with phenotypic information in a seamless process. Considered clinical data collection will be integral to this process.