An introduction to standardized clinical nomenclature for dysmorphic features: the Elements of Morphology project

Human structural malformations (anomalies, or birth defects) have a broad and complex range of manifestations and severities. The description of these findings can be challenging because the variation of the features is continuous and only some of them can be objectively assessed (that is, measured). Additionally, different specialties (embryologists, developmental biologists, surgeons, clinical geneticists, and so on) have developed descriptors for these findings that are based either on mechanistic, etiologic, or management considerations, and are thus laden with meaning, which can be invalidated by changing knowledge. Some terms have disparate definitions even within a specialty. Finally, a number of terms previously in common clinical use were disparaging (mongoloid slant, arachnodactyly, devil ear, and so on) and needed to be replaced with more neutral terms.

The need to develop such a set of terms is driven by the ever-increasing throughput of biological methods, which are outstripping the ability of clinical analyses to properly phenotype patients for both research and clinical care. The processes of the elucidation of the etiology of these disorders can be represented as a pipeline of varying caliber, where the diameter of the pipe represents the throughput of the process. In the late 1980 s and early 1990 s, the limiting factors were entirely molecular: it was much more difficult and slow to genotype and map disorders, find genes within candidate regions and sequence them than it was to identify and clinically analyze the patients (Figure 1a). With the completion of the human genome project, physical mapping could be performed by interrogating a web browser (Figure 1b). Soon after, high throughput capillary sequencing and single nucleotide polymorphism (SNP) marker typing methods improved to lessen those impediments. Finally, with the advent of chip-based genotyping with > 10⁶ features and next generation sequencing, molecular process throughputs are massive and no longer limiting (Figure 1c). Therefore, we have now reached a point where the ends of the process pipeline, which involves patient-related activities such as phenotyping and genotype-phenotype correlation, is the point limiting progress. The goal of implementing standardized terminology is therefore twofold: to increase the throughput of clinical analyses by developing a standardized common language that will allow development of databases for malformation phenotypes and to allow published case reports, case series, and genotype-phenotype correlation papers to serve as searchable repositories for clinical data. By harnessing these approaches, it is hoped that we can rapidly and effectively determine the etiology for hundreds or thousands of disorders and put these data to work, both as stimulants for basic science inquiries into mechanisms of normal and abnormal development and more directly for improved patient care.

The Elements of Morphology project was initiated to address the growing realization that the language used to describe human physical anomalies was inconsistent, incompletely defined, redundant, and in some cases confusing or even pejorative. On a practical level, these problems with the descriptive language are a barrier to text mining and database functionality [1]. To address this situation, an international group of clinicians resolved to develop a set of terms that could be used to describe human structural malformations. The attributes of this terminology set incorporated a number of important features [2]. Some of these features included; one-to-one correspondence of clinical terms to clinical manifestations, terms should be linked to appropriate, validated qualifiers, terms should not subsume multiple features if those features can occur alone, the terminologic set should be versioned so that it can be updated and referenced to prior versions. This effort was initiated in 2005 and resulted in a series of papers that described the rationale for the terms [3] and defined an initial list of terms in six categories (head and face [4], periorbital [5], nose [6], mouth and oral region [7], ear [8], and hands and feet [9]). Each of the terms was assigned a preferred descriptor, a one or two sentence definition, alternative objective and subjective definitions when appropriate, a list of acceptable synonyms, and a list of terms that are supplanted or replaced by the preferred term (see Figure 2 for examples). Definitions will be developed for another set of terms in the coming year. A web-based version of the terminology is also available http://​elementsofmorpho​logy.​nih.​gov/​. This website includes the full text and the photographs of all of the definitions, an index of terms and synonyms and a hypertext linked anatomical figure to facilitate rapid location of relevant terms, and a feedback and commentary site for each term that allows users to post queries and suggest corrections, modifications, or even additional terms.

Standards of terminology and reporting have been suggested; however, without good uptake by researchers and editors, confusion over terminology will persist. It was the intent of the group that the guidelines and suggestions would be taken up by the wider research community. These terms would be very useful for phenotype databases. The human phenotype ontology system [10] has begun to incorporate the Elements of Morphology terms and definitions into their system. An example of a practical implementation of the terms is the genotype-phenotype databases for the International Standards for Cytogenomic Arrays consortium (http://​iscaconsortium.​org/​ and see also [11]).

Going forward, it would be ideal for journal editors to determine that it is appropriate to ask authors of papers that describe malformations in humans to conform to this terminology. Efforts are underway to implement this terminology nomenclature for the American Journal of Medical Genetics and the American Journal of Human Genetics, in addition to the BMC series. The following approach is suggested for authors to follow, and editors to endorse for their publications:

Authors will need to use judgment in specifying features within tables. It is recognized that specifying, for example, an objective finding such as telecanthus could be cumbersome. A recommendation is that the paper should include a concise table (for example see Table 1) that simply lists 'telecanthus' or an indicator such as a '+' or 'tick' symbol for the finding but this table would link out to a supplementary data sheet for each patient that lists the findings (for example see Table 2).

The display of data in standardized formats, using defined terms, traceable to example images, commentaries, and norms, distinguishing absent findings from those not assessed, and distinguishing subjective from objective assessments, will allow readers and data miners to independently assess underlying data and conclusions and use case reports and other published clinical analyses to make new discoveries in the future. Like all new approaches to publishing and displaying data, these standards and approaches will need to evolve, and the BMC journal community, and hopefully all biomedical journals, should participate in this process. All of the editors look forward to feedback from the readers and authors on ways to improve and refine our approach to this challenge.