The article by Pandya et al.1 in this issue is a significant contribution to the field of medical genetics and will be especially helpful for those of us who work with children who have hearing losses. The article raises important issues with regard to the interpretation of DNA variation as it effects the phenotype: specifically, which mutations are pathogenic and how certain can the clinician be about a molecular diagnosis that implicates a specific gene?

The word “mutation” is often incorrectly used to imply pathology. Mutation has long been used to mean change in the DNA from consensus (or wild type). A mutation can be deleterious, neutral, or even beneficial. However, when one is said to be searching for mutations in a gene, it is usually interpreted to mean a search for pathologic variation. The determination that a particular change in the DNA is pathologic, or detrimental, depends upon a assessment of the impact of the mutation on the transcription and/or translation of the gene, as well as upon how the change might influence the actions of the protein being coded for by the gene. For a recessive disorder, this is not always a straightforward process. But there are rules that help guide one in making that decision. The two primary rules for excluding a mutation as being pathologic are (1) that the frequency of the mutant allele in a population must be below that which would be expected if everyone with the disease has that mutation and (2) that a normal phenotype can never occur in a person who is a true homozygote. For example, the frequency of Usher syndrome type I is about 3 to 5/100,000, say 5/100,000. The gene frequency of all the Usher type I alleles together cannot exceed 0.007; so, a “mutant” allele of any Usher I gene with a frequency in excess of that must, ipso facto, not be pathologic for that disease. No matter how severe, once a DNA change is observed as truly homozygous in an unaffected person, real questions must arise as to whether that mutation is pathogenic.

Any mutation that passes the first two criteria has a potential of being pathologic. Pathology can almost always be ascribed to a mutation that causes a stop, frameshift, or one that changes an invariant splice site. Rare missense mutations are often difficult to interpret. Arguments in favor of pathology include how well the involved protein domain is conserved across species and how the mutation is predicted to change the structure of the protein. Mutations that occur in an intron or that do not change any codon are usually accepted as not being pathologic, but our ignorance about how gene structure affects gene function creates lingering doubts: do these mutations effect regulation, do they affect mRNA stability, and are there other spliceforms that have not yet been recognized? For these reasons, Pandya et al.1 have been understandably cautious in ascribing pathology to some Cx26 mutations.

Once we accept a mutation as having a pathology, its presence in a patient still does not prove the patient’s hearing loss is due to a deficiency of that gene product. This depends upon the true genotype of the patient. One would think that after decades of use, the words homozygosity and heterozygosity would have clear meanings. Homozygous means simply that an individual has two copies of the same gene/allele/strand of DNA. For many patients, it is an inference and really means, “we saw only one mutation.” Heteroduplex, SSCP, sequencing, restriction fragment analysis, ARMS, etc., are techniques that are almost always blind to null alleles. So, a patient who is “homozygous” for 35delG may be a true homozygote or may be a compound heterozygote with 35delG as one allele and a null as the other. The null allele may be a deletion of all or part of that part of the gene being amplified, or it may be a change in the primer site for the amplification. Whichever the case, only one allele is actually being amplified and tested. In the absence of Southern hybridization, or dosage testing, the best evidence for a certain assignation of homozygosity comes from typing both parents, who would need to be heterozygous for the same mutation, in which case the inference of true homozygosity is warranted.

Heterozygote refers to the state where two identifiable alleles are present and this can usually be directly demonstrable in the laboratory. The term compound heterozygote usually refers to the situation where a person is heterozygous for two different but equally pathologic alleles. Heterozygous is also used to refer to individuals who have only one detectable pathologic mutation; although technically correct, it can be confusing in that it does not differentiate between a compound heterozygote and a heterozygote with a “hidden mutation.” An alternative might be to label such persons putative occult compound heterozygotes, or POCHs. A POCH may be nothing more than a heterozygote with one normal and one abnormal gene. In Cx26, 35delG occurs often enough, about 1% to 4% of the hearing population are heterozygotes, that even among people with hearing loss, one would expect to find a few cases of true heterozygosity where the conclusion that they have a CX26-based hearing impairment would be erroneous. But, realistically, the observation of only one pathologic mutation in a hearing-related gene, like CX26, considerably raises the probability that the hearing loss is due to that gene; the rarer the disease, the greater the likelihood. For example, a CX26 POCH will have a probability of about 90% of being a true CX26 compound heterozygote, with the hearing loss being directly due to a deficiency in Connexin 26; for the rarer Usher Ib, a MYO7A POCH will have over a 95% chance of being a true Usher Ib rather than a fortuitous “normal” heterozygote. These are important considerations for genetic counseling, and better methods for estimating true compound heterozygosity for two pathologic alleles versus normal heterozygosity need to be derived so that rates of error in assuming a specific genetic diagnosis can be estimated and taken into consideration while counseling.

It is interesting to consider to what extent POCHs represent missing, or occult, pathologic mutations. Are these mutations being missed because of technical problems with methods being used to scan the gene or are they occurring in regions not normally scanned (e.g., introns, regulatory segments, unrecognized spliceforms). The real value of identifying occult pathologic mutations is that their elucidation may reveal something new about the structure of the gene and/or its regulation. As such, POCHs are a legitimate, important, but unfortunately often ignored subject to study.