Table 1.
Odds ratio (OR) and 95% confidence intervals (CIs) for seropositivity to H. pyloria/CagA and risk of hepatobiliary cancers in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO).
| Serostatus | Total subjects | H. pylori seropositivitya | CagA seropositivity | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Case +/− | Control +/− | OR (95% CI) (unadjusted) | OR (95% CI) (fully adjusted)b | Case +/− | Control +/− | OR (95% CI) (unadjusted) | OR (95% CI) (fully adjusted)b | ||
| Hepatobiliary cancer | 536 | 92/87 | 162/195 | 1.31 (0.89, 1.92) | 1.14 (0.73, 1.79) | 71/108 | 94/263 | 1.96 (1.31, 2.94) | 1.96 (1.21, 3.18) |
| Biliary tract cancers | 221 | 35/39 | 53/94 | 1.64 (0.90, 2.99) | 1.76 (0.90, 3.46) | 25/49 | 32/115 | 1.96 (1.02, 3.76) | 2.16 (1.03, 4.50) |
| Gallbladder | 72 | 10/14 | 16/32 | 1.48 (0.51, 4.31) | 1.56 (0.43, 5.66) | 9/15 | 13/35 | 1.57 (0.57, 4.33) | 1.75 (0.49, 6.30) |
| Extrahepatic bile ducts | 75 | 14/11 | 19/31 | 2.17 (0.77, 6.09) | 2.94 (0.87, 9.89) | 7/18 | 11/39 | 1.40 (0.46, 4.29) | 1.55 (0.42, 5.68) |
| Ampulla of vater | 59 | 9 /11 | 16/23 | 1.13 (0.35, 3.70) | 1.71 (0.30, 9.82) | 7/13 | 8/31 | 3.39 (0.64, 17.93) | 10.86 (0.87, 135.49) |
| Liver cancer | 315 | 57/48 | 109/101 | 1.11 (0.68, 1.83) | 0.87 (0.46, 1.65) | 46/59 | 62/148 | 1.96 (1.17, 3.28) | 1.96 (0.98, 3.93) |
| Intrahepatic bile ducts | 39 | 5/8 | 9/17 | 1.18 (0.30, 4.63) | 1.15 (0.27, 4.96) | 5/8 | 4/22 | 2.91 (0.68, 12.45) | 3.14 (0.59, 16.73) |
| Hepatocellular carcinoma | 276 | 52/40 | 100/84 | 1.10 (0.65, 1.88) | 0.93 (0.45, 1.93) | 41/51 | 58/126 | 1.84 (1.07, 3.20) | 1.99 (0.88, 4.50) |
aH. pylori seropositivity is defined as positivity to ≥4 H. pylori antigens.
bFull adjustment included: cigarette smoking status (never, current, former), total cigarettes per day, alcohol consumption (<1 drink, 1–3 drinks, >3 drinks), college education, BMI (kg/m2), diabetes, gallbladder stones or inflammation, liver comorbidity (hepatitis or cirrhosis), hepatitis B, hepatitis C. Collinear variables and variables predicting outcomes perfectly were not included. Participants with missing data for variables adjusted for in multivariate models were included in the analyses with a flagged variable indicating missing data.