In attempting to diagnose a set of congenital disorders of glycosylation, a team from Sanford Burnham Medical Research Institute has highlighted some important pitfalls of using sequencing results to inform clinical diagnosis, specifically, the potential of somatic mosaicism to confound whole-exome and other broad sequencing techniques.
The report, published in the American Journal of Human Genetics last month, details the group's discovery of a causal gene in two male mosaic CDG cases. Using a standard filtering pipeline for whole-exome sequencing, researchers had not previously been able to identify the gene, SLC35A2, as the cause of disease in the children. But after doing Sanger sequencing analysis to look specifically for the mutation, the researchers found that the two affected males did indeed carry a causative mutation in SLC35A2 with a mosaic presentation.
Hudson Freeze, the study's senior author, told Clinical Sequencing News that the results indicate that exome sequencing alone may miss the genetic root of some cases of CDG — and potentially other diseases.
According to Freeze, the results suggest labs would do well to couple broad sequencing approaches with a sharp eye to biology and to biochemical measurements when analyzing such disorders.
"If you say it's X-linked and you say you're looking at a male, the prediction is that you are going to have only one copy, and so you should have 100 percent mutant allele," Freeze said.
"But if you have somatic mosaicism, as you filter [your exome-sequencing data] you say, 'well those numbers don't match.' Accurate diagnosis depends on what cells you take and the degree of representation of the allele and wild-type allele."
The takeaway, according to the researchers, is that where mosaicism is possible, biochemical data and biological understanding of disease can and should help narrow and augment analysis of whole-exome sequencing data.
In the study, Freeze and his colleagues were tipped off to the possibility that two CDG cases might have a mutation in SLC35A2 because of clinical and biochemical similarities to a third case. They had found that patient to be a mosaic for an SLC35A2 mutation using a Sanger sequencing approach and focusing on candidate genes chosen based on results from biochemical serum testing for the protein transferrin.
"This one transferrin pattern that came up in this first case in particular, that really led us to the idea that this ought to be a deficiency in a specific transporter … [which is what] it turned out to be," Freeze explained.
Then, the researchers turned to the two additional cases with similar biochemical profiles and clinical characteristics. Reviewing exome data from the University of Washington, the group found one case, a female with a missense variant predicted to be causative. For another male, the exome-sequencing data did not reveal a driving mutation.
However, because of the similarity of the three children's transferrin profiles, Freeze said his group decided to do a manual review of the data to look specifically at SLC35A2. They found that the male subject did indeed have a missense variant predicted to cause a deleterious substitution in the last transmembrane domain of SLC35A2.
According to the researchers, 9 percent of the SLC35A2 alleles identified at the variant site were reference alleles, suggesting somatic mosaicism. Sanger sequencing confirmed the result.
Freeze said that after sequencing the subject's parents, the group also confirmed that all the SLC35A2 mutations arose de novo.
"To have two patients as somatic mosaics in our small collection — only about 50 to 75 patients who are undiagnosed and going in for whole-exome sequencing — is phenomenal to me. That was not one bit expected," Freeze said.
The finding may have relevance for other disorders where researchers are discovering significant frequencies of de novo mutation, like other forms of intellectual disability, Freeze added. "If these spontaneous mutations are more common, then it seems to me that in that sort of setting somatic mosaicism is going to be much more frequent. So you have to be ready to find it."
"If you do just exome in the absence of some hardcore biochemical data, you really can get misled and you could miss things," he said.
Interestingly, the group also found that in the two mosaic male cases the initial aberrant transferrin profiles normalized later in life: at age three and age five, respectively. Since diagnosing the children's mutations was precipitated by this transferrin measurement — the mystery's "biological lead," according to the group — the researchers recommend that all individuals suspected of having a CDG should be tested for abnormal transferrin as early in life as possible.
According to Freeze, the study should not be taken to impinge the value of exome sequencing overall. CDGs, as a group, were virtually unknown before about the late 1990s, "and since that time," he said, "thanks in large part in the last few years to exome sequencing, there are now close to 90 different disorders known."
"But once you have a certain expectation and set your filters, you are going to get what you get," he said. "It was only because we had our other indications that we said 'let's look more closely,' and then when we Sanger sequenced, it was obvious that it was there."