The criteria developed and how the genes examined are assigned to one of three groups—namely, unlikely to be recovered (group 1), recoverability uncertain (group 2), and potentially recoverable (group 3)—summarized in Annex 4D. Since the starting assumption is that the genomic region containing the gene of interest has sustained a multigene deletion, the assessments only tell us which disease-causing mutations, if induced, may be recovered in live births within the framework of the criteria used; they do not shed light on the absolute radiation risk of a given genetic disease. Also worth mentioning here is that assignment to group 1 (unlikely to be recovered) is somewhat less subjective, and therefore more reliable, than that to the other two groups. This aspect is taken into account in defining PRCF (*i.e.*, by lumping groups 2 and 3 and considering that the mutations in the genes included may be potentially recoverable).

In general terms, if one analyzes a total of *N* genes and if *n* among them can be excluded as unlikely to be recovered, the remainder, made up of the other two groups, constitutes (*N − n*) and the fraction (*N − n*)/*N* provides a crude measure of genes at which induced mutations may be recoverable. This fraction is referred to as the unweighted PRCF.

The PRCF as estimated above, however, does not take into account differences in the prevalence of diseases assigned to different groups. For example, if a disease with a high prevalence is assigned to group 1, societal concern about radiation effects will be far less than when it is assigned to the other two groups. Consequently, some weighting for disease prevalence is required.

If *P* is the total prevalence of diseases due to mutations in *N* genes and *p* is the prevalence of (*N* − *n*) genes (in groups 2 + 3), then the weighted PRCF becomes [*p*(*N* – *n*)/(*PN*)]. For the purpose of risk estimation however, it is preferable to use a range provided by the unweighted and weighted PRCF estimates to avoid the impression of undue precision.

A total of 67 genes involved in autosomal dominant (59) or X-linked (8) recessive diseases was included in the analysis. The results, given in Table 4-5, show that the unweighted and weighted PRCFs for autosomal dominants are 0.29 and 0.16, respectively; when X-linked diseases are included, the corresponding values become 0.36 and 0.20. Since the overall estimated prevalence of autosomal dominants is an order of magnitude higher than that of X-linked diseases (*i.e.*, 1.5% versus 0.15%), the use of the range of (rounded) PRCF values of 0.15 to 0.30 (encompassing the prevalence-weighted and unweighted estimates) for autosomal dominant and X-linked diseases seems reasonable.

The recoverability of induced recessive mutations is also subject to constraints imposed by the structure, function, and genomic contexts of the underlying genes. However, since induced recessive mutations are first present in the heterozygous condition (and 50% of the gene product is sufficient for normal functioning), one can assume that even large deletions may be recoverable in heterozygotes (unless the induced deletion encompasses neighboring essential structural genes, resulting in inviability of heterozygotes). Additionally, induced recessive mutations, at least in the first several generations, do not result in recessive diseases, and as discussed earlier, the MC for recessive diseases is close to zero

TABLE 4-5 Assessment of Potential Recoverability of Radiation-Induced Mutations in Autosomal Dominant and X-Linked Diseases and Calculation of PRCFs