Table 4.

Variation within the IL8 gene and risk for GCA and ESCC

A
GenotypeGCC
ESCC
RR (95% CI)*P(2 df)P(1 df)RR (95% CI)P(2 df)P(1 df)
IL8 −251
 T/T1.00 (reference)0.0680.0481.00 (reference)0.360.57
 A/T1.11 (0.65-1.92)0.74 (0.47-1.18)
 A/A1.96 (1.03-3.75)0.97 (0.54-1.75)
IL8 +396
 T/T1.00 (reference)0.0220.0361.00 (reference)0.300.28
 G/T0.99 (0.57-1.72)0.76 (0.47-1.21)
 G/G2.06 (1.09-3.89)0.95 (0.52-1.73)
IL8 +781
 C/C1.00 (reference)0.170.0591.00 (reference)0.570.45
 C/T1.43 (0.84-2.43)0.85 (0.54-1.34)
 T/T1.76 (0.88-3.54)1.02 (0.55-1.87)
B
−251/+396/+781 Haplotype§Subcohort frequency (%)GCC
ESCC
RR (95% CI)P(5 df)RR (95% CI)P(5 df)
TTC/TTC135 (35.3)1.00 (reference)0.0341.00 (reference)0.61
TTC/AGT155 (40.5)1.25 (0.70-2.26)0.83 (0.50-1.37)
AGT/AGT57 (14.9)1.77 (0.84-3.72)1.02 (0.53-1.94)
TTC/AGC18 (4.7)0.29 (0.04-2.30)0.43 (0.12-1.63)
AGT/AGC9 (2.4)4.14 (1.31-13.1)0.67 (0.14-3.25)
TTC/ATT9 (2.4)1.94 (0.48-7.78)0.70 (0.14-3.43)
C
−251/+396 Haplotype§Subcohort frequency (%)RR (95% CI)P(3 df)RR (95% CI)P(3 df)
TT/TT143 (31.9)1.00 (reference)0.0191.00 (reference)0.14
TT/AG217 (48.4)0.96 (0.55-1.67)0.65 (0.41-1.04)
AG/AG73 (16.3)2.12 (1.11-4.04)1.02 (0.56-1.84)
TT/AT15 (3.4)2.15 (0.71-6.48)1.43 (0.50-4.07)
  • * RRs and 95% CIs were generated from models stratified on age and sex with additional adjustment from continuous age variables for each strata and variables for smoking, drinking, and trial.

  • This P value comes from the score test for the addition of all the genotype/haplotype indicator variables to the base model simultaneously. It assesses the overall association between genotype/haplotype and cancer. 2, 3, or 5 df denote a P global.

  • This P value comes from the score test for the addition on a single variable for genotype, where Wt = 0, Hz = 1, and Hv = 2. 1 df denotes a P trend.

  • § Two haplotypes were inferred for each subject using PHASE software (see Materials and Methods). Haplotypes were constructed only if genotype data were available at all loci for each individual. Individuals with two copies of the most frequent haplotype were set as the reference group and the RRs and 95% CIs associated with other diploid states were estimated using the same models as described in footnote *.