genomics & society

Whole Exome Sequencing: The NIH Experience

Posted by Myra I. Roche on November 28, 2011

In a new post, CGS trainee, Dragana Lassiter, summarizes the discussion of the November, 2011 CGS Seminar entitiled: “Gifts of the Body: Expectations of Cancer Patients Involved in a Whole Genome Sequencing (WGS) Study” that was presented by CGS trainees, Rachel Haase and Marsha Michie along with with CGS investigator, Debra Skinner.   Dragana discusses how the anthropological perspective of the meaning of gifts can provide a way to explain why individuals may consent to participate in research studies involving sample donation.  Many centers are now enrolling patients in studies that sequence either the whole genome (WGS), the exome (largely the coding sequences) (WES) or use a combination of approaches including SNP arrays.   But what has been learned about the process, so far? In a recent paper in Genetics in Medicine, found here, the NIH experience is summarized.  In it, the authors describe their process of patient selection and the strategies they used to handle the large amounts of data generated.

The Diagnostic Genotyping Experience at NIH
While some molecular genetics laboratories, such as Baylor College of Medicine, have just begun to offer WES clinically, the Undiagnosed Disease Program (UDP) at NIH has been using a combination of SNP arrays and WES  for the last couple of years.  Their focus, as the name suggests, is on making a diagnosis in symptomatic  individuals.   The goal of their program is to “achieve a diagnosis” for patients who had previously undergone an “exhaustive ” work-up but remained enigmatic.  A second aim of the project is to discover “new disorders ” in order to gain an insight into disease mechanisms.

Enriching the Sample for Success
The article details the extensive filtering process of both patients and data that occurred.  Of the 1191 medical records reviewed for potential candidates, only 326 patients were “accepted” and 160 actually enrolled.  The value of enriching the samples for those likely to have an identifiable mutation was underscored by the successful diagnosis in 39 cases; a respectable 24% success rate considering that these patients had already been extensively evaluated by the state of the art technologies.   It should be noted, however, that several clinical diagnoses were of  “common” conditions of adulthood that are not traditionally considered to be genetic, such as fibromyalgia.

Diagnostic Importance of Clinical and Family History Information
Interestingly, 12 of the 39 patients, nearly one-third, were diagnosed the old-fashioned way; by clinical features with no NextGen technology needed.  This low tech method diagnosed four times the number (12 vs. 3) that were diagnosed by WES.  Since all disorders were rare and some even labeled as “ultra rare” , it should not be particularly surprising that these patients had been on a diagnostic odyssey for many years.   But while technological advances receive a lot of press (hype?) and have, unquestionably, made a significant impact on the ability to make diagnoses where none could have been made before, these augment, not replace, the need for the clinical component.  The components of a comprehensive clinical genetics evaluation are: 1) a careful and accurate description of the patient’s features, 2) a comprehensive medical and family history, and 3) correct interpretation of (prior) laboratory data.  These elements are needed, regardless of whether the evaluation is for intellectual disabilities in children, autism, or cancer.

Genomes in Common
The finding that affected individuals in the study population were much more likely to have large regions of homozygosity, as identified by SNP array testing, than the controls underscores the need to expect (and counsel patients about) the unexpected.  While multiple small regions of contiguous homozygosity are common in the general population, very long regions imply that the parents shared some degree of relatedness and the individual inherited the same genomic region from a common ancestor.  Extended regions of homozygosity have been a very useful tool allowing researchers to identify rare recessive risk variants that contribute to the disease in the family.   Many examples exist including that for Charcot-Marie-Tooth disease, found here.   But this is the kind of incidental findings that most individuals being tested would never suspect could be identified but yet has serious social, ethical, legal and counseling implications.

Finding the Unexpected
The unexpected results of a group of patients likely conceived through a “first-degree” familial relationships (father-daughter, mother-son, brother-sister) is reported here by Schaaf, et. al.   As the authors’ state:  “In cases where the mother is a minor, clinicians who uncover a likely incestuous relationship could be legally required to report it to child protection services and, potentially, the police, since the pregnancy might have occurred in the setting of sexual abuse—perhaps by a father or brother. The physician’s duty to report is less clear in cases where the mother is an adult and might depend on whether she was a minor or an adult at the time she became pregnant.”   They recommend counseling protocols be expanded to include a discussion of  the possibility of uncovering undisclosed, and possibly previously unknown, consanguinity.  They further note that intellectual disabilities are known to be frequent in children born of “incestuous parentage” further complicating the issues.

Which Conditions Were “Sequence-Worthy?”
The most common phenotypic categories accepted for testing by the UDP were neurological and psychiatric disorders.  This selection underscores the historic difficulty in clinically classifying these conditions into discrete etiological classes.  While many were called and few were chosen, affected children were chosen much more frequently than affected adults.  Their acceptance rate into the program was 47% vs. 20%  for adults.  Two “new” disorders were identified but in an interesting twist, the authors listed the phenotypes of 10 individuals who are still searching for their diagnosis.  This “have you seen me before” approach, reminiscent of the face on the milk carton appeal, attempts to harness the collective wisdom and experience of their readers, many of whom are likely to be geneticists.

How to Handle the Data
As with the winnowing down of the number of patients, the authors describe a similar process of winnowing down the data to a more manageable set.  One powerful technique was to sample informative family members for comparison, a process that “often decreased the number of final candidate sequence variants by 1-2 orders of magnitude”.   A second strategy was to use databases such as dbSNP.   The combination of filters often reduced the number of candidate genes to 20 or fewer which allowed extended analysis to be done on just a handful of genes rather than thousands.

The Phenome
In the end, it was the need for “accurate and meticulous phenotyping” by both physical exam and targeted diagnostic testing that arose as the most serious bottleneck in the process.  The descriptions from the medical records that accompanied the enrolled patients were often quite strikingly discrepant from the actual features of the person sitting on the exam table.  This plea for accurate clinical documentation may seem ironic embedded in an article about the use of the most current technology but classifying phenotypes is now considered the “rate-limiting step in advancing knowledge about human disease according to “Bilder, et. al.  In their 2009 article,  they described the relatively new “ome” on the block, phenomics, as the “systematic study of phenotypes on a genome-wide scale” that attempts to link genetic variation to public health problems.  They also compared phenomics with its relatively simpler cousin, genomics:  “The human genome, with only three billion bases, selected from a pool of only four nucleic acids, organized in a neat one-dimensional sequence, pales in comparison to the human phenome, which contains an unknown number of elements, many of which are characterized by enormous inter-individual variation that is at best only partially understood, and for which the dimensionality remains unknown.”  As the cost of sequencing continuous to plummet, the capacity to store, interpret, and communicate the data lag farther and farther behind.


2 Responses to “Whole Exome Sequencing: The NIH Experience”

  1. Marsha Michie said

    Myra, you make an excellent point that clinical observation and rigorous phenotyping is still perhaps the most critical component of diagnosis, even for genetic disorders. Given the continuing squeeze on health care providers’ time, lengthy interactions with patients now seem to be a luxury most of the time. Do you see the introduction of WES into the clinic as furthering this trend, in that physicians (despite your warnings) may see genetic testing as an increasingly inexpensive “shortcut” to bypass time-intensive personal interactions?

    Also, in looking at the article on the NIH UDP program, I was struck by the utilization of DNA from family members in attempting to diagnose rare diseases via whole genome/exome sequencing. While family members may have been especially motivated to help out in these cases, where a long diagnostic odyssey had not been fruitful, how feasible is it to rely upon family members to “pitch in” with their DNA in this way? I would be interested to know how important this kind of within-family comparison will be for WES/WGS diagnosis, and how willing family members are to contribute to that effort. Except in the case of parents of children with a serious disorder, this kind of request for family DNA seems to run counter to the overall “individualist” paradigm of medicine–particularly in the U.S. where adults are expected to take individual responsibility for their own health.

  2. Myra I. Roche said

    Thanks for your comment, Marsha. I think that the constraining elements preventing many physicians from ordering genetic tests, are 1) they often lie outside their radar and 2) they understand that the post-test counseling requires more expertise than they have and more time than they can afford. Jumping to a genetic test is really not a shortcut because of the many issues involved in interpreting even a straightforward genetic test result (e.g. implications for reproductive risks, options, implications for other relatives, etc..) . Genetic test results are rarely definitively interpretable without a well-defined phenotype and many physicians simply do not have the skill set to accurately describe a genetic phenotype.

    As for the family studies, being able to analyze DNA from triads (both parents and affected child) allows the dismissal of some genetic changes that are simply familial, benign duplications, say, that are inherited but not pathogenic. This is a very important filter that helps sift out benign variants. In some cases, siblings could also be informative. The genetic information of more distant relatives would usually be less helpful in determining if a variant of unknown significance was pathogenic or not.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s