NEW YORK (GenomeWeb) – Putative loss-of-function mutations may be linked to phenotypes, including disease phenotypes, more often in people than previously thought, according to a new study from researchers at the National Human Genome Research Institute.
Through iterative phenotyping of a ClinSeq cohort, NHGRI's Leslie Biesecker and his colleagues found that about one in 30 people have a putative loss-of-function mutation that's linked with a phenotype in either themselves or in a family member, findings they reported in the American Journal of Human Genetics today.
"Our estimate of 1/30 provides a good starting point for the aggregate probability of a phenotype attributable to a heterozygous LOF variant," Biesecker and his colleagues wrote.
The researchers recruited 951 ClinSeq participants and collected initial family history information through the US Surgeon General's My Family Health Portrait tool. A genetic counselor or geneticist then went over that information with the participants to confirm the diagnoses, ask about other diseases, and expand pedigree information to three generations. The cohort was selected for a range of atherosclerosis phenotypes, the researchers noted.
They also collected DNA samples from the participants for exome sequencing, and generated nearly 2.5 million variants, of which 13,903 were designated as putative loss-of-function mutations. After filtering for pLOF variants in genes where they'd likely cause a phenotype in a heterozygote, the researchers came up with a list of 82 variants in 103 participants.
Biesecker and his colleagues then attempted to re-contact those 103 participants. As 24 people were either lost to follow-up or declined to participate further, they were left with 79 participants with 64 pLOF variants of interest.
More than 43 percent of the participants had a finding or family history that could be due to the variants. After adjusting for atherosclerosis ascertainment bias, the researchers estimated a prevalence of one in 30, which they noted was a conservative estimate.
For instance, one person with a mutation in the SLC4A1 gene had both a personal and familial history of spherocytosis, while another person with a mutation in PKD1 had a personal and familial history of polycystic kidney disease.
But at the same time, they found no indication of cancer in two participants with variants in the BRCA1 gene or in their families, and no sign of Duchenne muscular dystrophy in one participant with a DMD gene variant.
Forty-two of these participants went through a second round of phenotyping, either at NIH or elsewhere, and of these, 19 individuals were positive for phenotypes associated with their variant of interest based on examinations, biochemical findings, and other tests.
For example, a participant with a MYH7 variant was found to have left ventricular non-compaction based on echocardiogram and MRI results, and another participant with a mutation in KCNQ4 was discovered to have some hearing loss based on a formal hearing test.
Meanwhile, 21 individuals with variants in 18 difference genes who underwent a second round of phenotyping were negative for associated phenotypes upon examination. Four people with MYOC variations, which are predicted to cause glaucoma, had normal eye exams and intraocular pressures, and another participant with an APT2C1 variant had a normal skin examination. A targeted family history also revealed no links to disease in these participants.
The researchers noted that though their predictive positive rate was high at 43 percent, it was not perfect. Still, as there is no other way currently to predict the presence of a phenotype for pLOF variants, they argued that researchers and clinicians would have to adopt this sort of iterative phenotyping approach to evaluate people with novel or unexpected pLOF variants.
They added that their data indicated that individuals were at higher risk of an autosomal dominant disorder than previously thought, as they uncovered 28 participants out of the original 951 that had a rare, pLOF variant — or about one in 30 — which they noted was an increase from previous estimates of between one in 500 to one in 500,000.
"This change in risk is enormous and can be used by a thoughtful clinician to prompt further evaluations of the proband and their family members for evidence of a discernable disorder," the researchers added. "Although it is true that not all of the phenotypes detected here are medically actionable, this study serves as a proof of principle that there might indeed be predictive value in healthy genomes and exomes."