Fluidos de perforación

Genes appear important also in cortical cataract, with a heritability o f 53 to 58% in this population, and this inheritance appears to involve dominant genes (Table 39). Unique environment explained 26 to 37% o f the variance. These figures compare with the heritability o f 48% for nuclear cataract which had a lesser environmental effect o f only 14% compared to cortical cataract. Age effects were more important in nuclear cataract, explaining 38% o f the variance compared to 11-14% o f the variance o f cortical cataract.

These figures are interesting, suggesting that cortical cataract may in part be dominantly inherited. The previous family study o f cortical cataract using

commingling analysis showed two transformed distributions fitted better than one,’"" which would fit with a dominant transmission hypothesis (or a recessive hypothesis). However, complex segregation analysis predicted a major recessive gene accounting for 45% o f the variance o f women. There is an extensive set o f assumptions in

complex segregation analysis, and violations o f assumptions (eg skewness and kurtosis) can mimic the effects o f a major locus. Twin studies do not estimate the number o f genes involved, and assume a multifactorial causality (genes and environment) with underlying liability distribution.

Model fitting analysis o f both methods o f grading suggested dominant genes are important in cortical cataract inheritance (Table 38), reflected in the significant loss o f fit when D, the effect o f dominance, was dropped from the modelling. As discussed earlier, twin studies have low power to detect dominance due to the low DZ

correlation,^^® which explains the wide confidence intervals. Genetic models assume the effect o f dominance is additional to an additive genetic effect,^^^ so the effect o f removing additive genes from the model cannot be tested, even though the confidence intervals for estimation o f A cross zero in the cortical cataract data.(Table 39)

4.2.4 Age-related macular degeneration

The low concordance between twins for the early features o f ARM in this population sample is surprising, given the smaller series o f twins which have shown a greater similarity. Meyers included in his definition o f macular degeneration individuals with more than 20 hard drusen, and his 25/25 MZ concordance and 5/12 DZ concordance (o f 134 twin pairs examined)^^ does not mention how concordant the twins were. This means that if one twin had only hard drusen and the other exudative AMD, they were classified as concordant. However, he did note concordance o f exudative AMD in 4 MZ twin pairs and all 15 MZ pairs were concordant for non-exudative disease. The study may also be criticised as there were 98 MZ pairs and 38 DZ pairs

examined, certainly not representative o f the population, and recruitment at twin fairs m ay result in recruitment o f twins more concordant as they hold a greater “twin identity”. It may be that the mean age o f our twins (62) means that only early disease was seen and that the twins may become more concordant with time - a ten year follow up would be very interesting. In fact M eyers’ twins had a mean age o f 64.9 for the MZ twins and 62 for the DZ twins, not dissimilar to this study.

Population-based twin studies always have a lower concordance than more selected twin samples, and often require huge numbers o f twins to gain sufficient statistical power for calculation o f heritability, particularly for binary traits. As an example, a recent pooling o f all the Scandinavian twin and cancer registries analysed over 44 000 twins to estimate the heritability o f cancers. For breast cancer, there were only 42 concordant M Z twin pairs and 505 discordant, and for DZ twins 52 concordant pairs and 1026 discordant. This resulted in an estimate o f the heritability o f breast cancer o f 27% with 95% confidence intervals o f 4 to 41%, which are wide despite the large numbers involved.^^^

Heiba attempted to improve power in his family segregation study from the Beaver Dam Eye Study by grading all individuals on a 15-point categorical scale, including all phenotypes from 1 hard druse through to soft drusen through to disciform

scarring.^^^ We decided not to do this with the twin study, as this form o f categorisation is not supported biologically: patients do not progress through the stages and so they do not reflect a true categorical scoring - soft drusen (level 6) may

occur without hard drusen (level 4) and who is to say that small hard drusen with pigmentary changes (level 9) are “more severe” than soft indistict drusen (level 7)?

The confidence intervals o f the heritability estimates in this study are wide, reflecting the lack o f power o f a bivariate statistic, even in a study o f this size. Despite the concerns about the power o f this study o f ARM, the parameter estimates that were significant suggest that genes are important in the heritability o f soft drusen (67% heritability -Table 45) but not so for pigmentary changes (estimate o f heritability 38% but 95% confidence intervals almost include 0% for genetic effect and 100% for environmental effect). This suggests that for candidate gene studies o f AMD, more attention should be paid to the phenotype o f soft drusen >=125p size rather than pigmentary changes or indeed few hard drusen. These phenotypes seem much more likely to be environmentally mediated. In addition, more than 20 hard drusen appear genetic and dominantly inherited, with a heritability o f 83%. The gene for dominant drusen associated with Malattia-Levinese/Doynes Honeycomb Retinal Dystrophy has been identified.^^^ This gene was examined as a candidate gene in 494 individuals with AMD and none had a mutation. However, the definition o f AMD was not

discussed in the published study, so we do not know whether the phenotype o f AMD included multiple hard drusen, but clearly this particular gene is not involved in the aetiology o f AMD.