Introduction
Disease-based epidemiological studies provide estimates of the minimum population prevalence of mtDNA disease as ∼1 in 5000,1–6 but the incidence of new mutations and the frequency of asymptomatic carriers have not been fully established. Fundamental differences between mtDNA inheritance and the Mendelian system mean that de novo mutation rates and carrier frequencies cannot be deduced with a standard Mendelian population-genetic approach. Strict maternal inheritance results in negligible intermolecular recombination apparent at the population level, and the presence of thousands of copies of mtDNA within each diploid mammalian cell adds a further complexity.7 Most pathogenic mtDNA mutations are heteroplasmic, with varying amounts of mutated mtDNA present within each cell.8 Phenotypic expression is ultimately dependent upon the proportion of mutated mtDNA or the amount of wild-type mtDNA within vulnerable tissues. This only leads to a biochemical defect of the respiratory chain when a critical threshold level is exceeded, with the precise amount varying from mutation to mutation.8 In keeping with this, a family with mitochondrial disease usually presents clinically for the first time when an individual inherits a high proportion of mutated mtDNA from a mother who harbors a low level of mutated mtDNA and remains asymptomatic. Such dramatic changes in heteroplasmy can occur in a single generation because of a restriction in the intracellular mtDNA content during embryonic development that is responsible for the mtDNA genetic bottleneck.9
Previous estimates of the carrier frequency of mtDNA mutations were largely based on the ascertainment of pedigrees through one or more clinically affected individuals, followed by careful family tracing of the maternal lineage.1–6 This raises the possibility that de novo mtDNA mutations remain undetected in the population because they are well below the threshold required for phenotypic expression or because they lie below the detection threshold of conventional sequencing. Many of these mutations would be lost through random genetic drift, causing disease only when the percentage level of mutation drifts upwards in one or more family members and exceeds the threshold required for clinical presentation.
A further complication is the strong association between specific pathogenic mtDNA mutations and closely related mtDNA polymorphisms (haplogroups), which is not thought to be due to a founder effect.10,11 However, it is not known whether the polymorphisms that define the haplogroup predispose it to de novo mutation events, whether they cause the preferential segregation of pathogenic mutations, or whether they affect clinical expression of the associated disorder. The only way of addressing these issues is by screening a large random sample of the population, with matched maternal samples, to determine whether the detected mutations are de novo or inherited.