Table 3

Epistasis between P2RX7 1405G and HLA DR3 in anti Ro±La seropositive primary Sjögren's syndrome (pSS)

Seropositive pSS (n = 95) versus controls (n = 136)

Seropositive pSS (n = 199) versus controls (n = 534)

Seropositive (n = 191) versus seronegative pSS (n = 80)

Seronegative pSS (n = 82) versus controls (n = 534)

Cohort 1: DRB1-typing

Cohort 2: DR3 proxy SNPa

Cohort 2: DR3 SSPa

Cohort 2: DR3 proxy SNPa

Term

β

SE

P

β

SE

P

β

SE

P

β

SE

P


β0 (Intercept 1)

-1.64

0.28

<0.001

-0.01

2.64

1

-3.97

4.31

0.36

1.04

3.32

0.75

β1 (DR3-pos)

2.23

0.37

<0.001

1.35

0.21

<0.001

1.39

0.37

<0.001

0.28

0.33

0.40

β2 (P2RX7_1405G)

1.11

0.38

0.003

0.26

0.22

0.24

0.25

0.34

0.46

0.29

0.27

0.28

β3 (DR3*1405G)

-1.27

0.53

0.016

-0.76

0.35

0.027

-1.27

0.56

0.023

0.02

0.47

0.96


aAnalysis performed using principle components adjustment for possible ancestry differences. The combined effect of these alleles was examined by two-factor interaction logistic regression (DR3*1405G) in both cohorts for the data tabulated in Additional File 1, Table S3. DR3 was coded positive (pos) or negative, and additive genetic coding was used for 1405G. The regression coefficients β1 and β2 represent the loge odds ratios for individuals who carry DR3 alone and 1405G alone, relative to individuals who carry neither. The interaction term, β3, represents the deviation from the expected log odds ratio for individuals who carry both alleles, and is the test for epistasis between 1405G and DR3. Different analyses were required to accommodate differences in the available data for DR3 status. The results demonstrate that DR3 (β2), or proxy, is a significant risk factor for seropositive pSS, but not seronegative pSS. Further, there is a negative epistatic interaction between DR3 (or proxy) and P2RX7 1405G (β3) specifically in seropositive pSS. However, the association of the 1405G allele with seropositive pSS (in the absence of DR3, β2), observed in Cohort 1, was not replicated in Cohort 2. SE, standard error.

Lester et al. Arthritis Research & Therapy 2013 15:R71   doi:10.1186/ar4248

Open Data