We often think of rare diseases in simplistic terms — a single gene in which mutations cause a recognizable disorder. That disorder might affect multiple cells, tissues, and organs, and the severity can vary significantly among patients. When a diagnosis is obtained from genetic testing, it’s tempting to consider all aspects of the patient’s condition as deriving from that disorder.
After all, if a patient already has one rare genetic disease, how likely is it that they have a second one? In other words, how often does lightning strike twice?
According to a Baylor-led study published in The New England Journal of Medicine last year, it’s around 5%.
Clinical Exome Sequencing
Jennifer Posey and colleagues retrospectively analyzed data from 7,374 consecutive unrelated patients referred for a clinical lab for “exome sequencing”, which sequences the coding regions of all genes (around 3% of the genome). The period for the study was October 2011 to April 2016, which suggests that Baylor was sequencing around 140 cases per month. That’s a massive clinical sequencing operation. As a point of comparison, a major children’s hospital such as ours might refer 150-200 patients per year for clinical exome sequencing.
Of the 7,374 cases sequenced, 2,076 (28.2%) received a molecular diagnosis. This is a typical diagnostic rate for clinical exome sequencing and is consistent with previous reports based on smaller cohorts. It’s possible to achieve a higher (or lower) “hit rate” by pre-selecting patients to sequence, but this large dataset provides some reassurance of the ~25-30% baseline.
Cases with Multiple Diagnoses
Of the 2,076 patients for whom clinical exome sequencing was informative, 101 (4.9%) had multiple molecular diagnoses:
- 97 had two molecular diagnoses
- 3 had three molecular diagnoses
- 1 had four molecular diagnoses (!)
The authors found no correlation between multiple diagnoses and age or sex. Most of these cases were referred by medical genetics (81.2%), but that specialty is responsible for the vast majority of exome referrals. A small portion of patients (6/101) had one known molecular diagnosis at the time they were referred for sequencing. In other words, the clinician suspected (correctly) that something else was going on.
I’d be very curious to know the total number of patients who were sent for exome sequencing even though they had one molecular diagnosis already. That would give us an idea of the success rate for finding a second diagnosis when one is suspected.
Forms of Multiple Genetic Diagnoses
The 101 multiply-diagnosed patients carried a total of 207 molecular diagnoses. Of those, more than half (112, or 54.1%) were due to pathogenic variants in autosomal dominant disease genes. Autosomal recessive genes accounted for another 30% of diagnoses, while X-linked disease genes (on the X-chromosome, if that wasn’t obvious) accounted for the remaining 16%.
The three most common combinations of molecular diagnoses involved autosomal dominant disease genes:
- 34% of patients had two pathogenic variants in two (different) AD genes
- 26% of patients had one pathogenic AD variant and one autosomal recessive (AR) diagnosis
- 16% of patients had one pathogenic AD variant and one X-linked (XL) diagnosis
The Role of de novo Mutations
In recent years, reports from several clinical sequencing laboratories have suggested that a large fraction of successfully diagnosed cases have a de novo mutation (present in the patient, absent from both parents) in a dominant disease gene.
This trend is further supported by the NEJM study of multiple diagnoses. The 101 patients had 207 molecular diagnoses, and de novo mutations accounted for:
- 68% of pathogenic variants in autosomal dominant (AD) disease genes
- 51% of pathogenic variants in X-linked (XL) disease genes
In other words, most of the pathogenic variants in AD and XL genes (which together accounted for about 70% of diagnoses) turned out to be de novo. This is almost certainly an underestimate, since the inheritance could not be determined for ~10% of variants in AD genes (probably because parental samples weren’t available). Also, every patient with a dominant or X-linked diagnosis had at least one de novo mutation.
An interesting corollary here is that around 14% of diagnoses involved an autosomal dominant disease gene with a variant inherited from a parent. Although it’s not explicitly stated here, in most of these studies, both parents are presumed to be healthy. In this subset, either that’s not true or the variant is not fully penetrant.
The Role of Homozygosity
Some 42 patients who received an autosomal recessive (AR) diagnosis were sequenced as part of a family trio, meaning that both parents’ genotypes were available. This involved 53 variants in total, about half of which were homozygous in the patient and heterozygous in both parents. This is classic autosomal recessive inheritance: the 1-in-4 chance of receiving the “defective” copy from both parents (you remember Punnett squares, don’t you?).
Some 22 patients with a homozygous recessive variant were also noted to have one or more regions of “absence of heterozygosity” in their genomes. This could be evidence of a larger genomic alteration, such as uniparental disomy (UPD, meaning that you get two copies of one parent’s chromosome rather than one copy from each). More often, however, it’s because the parents are distantly (or not so distantly) related.
Often, we know this in advance, either because the parents know they’re related, or a SNP array revealed less than the expected amount of heterozygosity (variability between chromosomes). In this study, the authors reviewed SNP array data for 27 samples that had a recessive diagnosis, and found that 22 of them had one or more regions of reduced heterozygosity (ROH) larger than 10 Mbp. For 15 of these 22 patients, consanguinity (parental relatedness) was reported.
Absence or reduced heterozygosity means that a patient’s genome contains long stretches of homozygosity, which can be problematic if that region harbors a variant in an autosomal recessive disease gene. Thus, patients with known — or suspected — consanguinity are carefully scrutinized for autosomal recessive disease genes. I refer to “suspected” here because consanguineous unions are more common in certain regions of the world (e.g. the Middle East) or among certain cultural groups (e.g. Amish, Mennonite, and Ashkenzi Jewish).
Not Just Homozygous Variants
Among the 22 patients with ROH in this study, homozygous variants in autosomal recessive genes accounted for the majority (66%) of diagnoses. That’s the expected finding. The unexpected finding was that a minority of them had diagnoses of variants that fit other inheritance patterns (5 cases with autosomal dominant variants and one with an X-linked variant).
It’s interesting, because the smart money would have been on homozygous variants. Yet another quirk of human genetics that we can file under “it’s complicated.”