The distribution of
PRPH2 (NM_000322.4) haplotypes in
trans for PD, CRD, CACD, and RP phenotypes are shown in
Tables 3 and
4. For individuals over 43 years, in Group I, representing milder disease with different types of PDs, Gln
304-Lys
310-Asp
338 (C
910-A
929-A
1014) was identified in eight individuals (29.6% vs. 0%) and 19 had Glu
304-Lys
310-Gly
338 (G
910-A
929-G
1013). In Group II, with CRD, RP, and CACD, 17 individuals had Glu
304-Lys
310-Gly
338 (G
910-A
929-G
1013; 94.4% vs. 70.4%), and one had Gln
304-Arg
310-Asp
338 (C
910-G
929-A
1013). Gln
304-Lys
310-Asp
338 (C
910-A
929-A
1013) was absent (
Table 4). A significant difference in the haplotype distribution between the two phenotypic groups among individuals older than 43 years was seen using Fisher's exact probability test (
P = 0.0092). A GEE analysis that accounted for differences in age and familial correlation of genotypes confirmed that the increased frequency of the Glu
304-Lys
310-Gly
338 haplotype among individuals with CRD, RP, and CACD was significant with a relevant
P value of <2e-16. However, because the asymptotic
P value may not be robust to the “zero” cell counts in two of the haplotype-phenotype categories (see
Tables 3,
4), we reran the GEE model taking the conservative approach of collapsing the Gln
304-Lys
310-Asp
338 (C-A-A) and Gln
304-Arg
310-Asp
338 (C-G-A) haplotypes into a single category (
P = 0.03,
N = 45 due to removal of individuals with age <43). The estimated odds ratio for this comparison is 7.16 (95% CI = [2.8–18.4]), suggesting that the odds of having the CACD/RP-like phenotype (versus the PD phenotype) are 7.16 times greater for affected individuals with a Glu
304-Lys
310-Gly
338 haplotype (G-A-G) haplotype in
trans than for affected individuals with one of the other two haplotypes in
trans (Gln
304-Lys
310-Asp
338 [C-A-A] and Gln
304-Arg
310-Asp
338 [C-G-A]).