As shown in
Table 5, both mean perifoveal drusen area and volume increased as the
CFH score increased, and the trends for increasing drusen burden as the
CFH score increased were significant after adjusting for age, sex, education, smoking, BMI, and AMD grades (
P trend = 0.004 for area and
P trend = 0.002 for volume). Carriers of two risk alleles versus zero or one had higher drusen area (
P = 0.03) and drusen volume (
P = 0.04), and carriers of three or four risk alleles also had significantly higher drusen area (
P < 0.001) and volume (
P < 0.001) compared with having zero or one risk alleles. Similar comparisons were assessed for the
ARMS2/HTRA1 variant: mean drusen area and volume increased as the number of risk alleles increased, after adjustment for other variables (
P trend < 0.001 for both drusen area and volume). Carrying two risk alleles versus none was significantly related to higher drusen area and volume (
P = 0.008 and 0.004, respectively). These associations between OCT-derived perifoveal drusen measurements and genetic factors are also depicted in
Figure 3.
When both genes were adjusted for each other (bivariate analyses shown in
Table 5), the associations between drusen measurements and
CFH score were somewhat reduced. However, the trend for higher drusen area and volume with higher score remained significant (
P trend = 0.01 for area;
P trend = 0.005 for volume). When the
ARMS2/HTRA1 genetic variant was adjusted for the
CFH score, results were essentially unchanged from the univariate analysis as above, and trends for increasing drusen area and volume with increasing number of
ARMS2/HTRA1 risk alleles were significant (
P trend < 0.001 and 0.001 for area and volume, respectively).
All tests of interaction between CFH score and ARMS2/HTRA1 were not significant. When comparing drusen area or volume measurements ≥median versus no measurable drusen, the P values were 0.38 for drusen area and 0.33 for drusen volume (data not shown).