Table 3.
The amount of Fusobacterium nucleatum in colorectal carcinoma tissue and the density of T-cells
| Univariable OR (95% CI) |
Multivariable OR (95% CI)a |
|||
|---|---|---|---|---|
| Model for CD3+ cell density (n = 567, as an outcome variable) | ||||
| The amount of Fusobacterium nucleatum |  |
Negative | 1 (reference) | 1 (reference) |
| Low | 0.68 (0.37–1.25) | 0.63 (0.34–1.17) | ||
| High | 0.50 (0.27–0.91) | 0.47 (0.26–0.87) | ||
| Ptrendb | 0.012 | 0.006 | ||
| Model for CD8+ cell density (n = 560, as an outcome variable) | ||||
| The amount of Fusobacterium nucleatum | |
Negative | 1 (reference) | 1 (reference) |
| Low | 0.60 (0.32–1.12) | 0.66 (0.35–1.23) | ||
| High | 0.71 (0.39–1.30) | 0.79 (0.43–1.44) | ||
| Ptrendb | 0.11 | 0.24 | ||
| Model for CD45RO+ cell density (n = 574, as an outcome variable) | ||||
| The amount of Fusobacterium nucleatum | |
Negative | 1 (reference) | 1 (reference) |
| Low | 1.33 (0.72–2.43) | 1.25 (0.67–2.31) | ||
| High | 1.09 (0.60–1.98) | 0.97 (0.53–1.79) | ||
| Ptrendb | 0.56 | 0.88 | ||
| Model for FOXP3+ cell density (n = 548, as an outcome variable) | ||||
| The amount of Fusobacterium nucleatum | |
Negative | 1 (reference) | 1 (reference) |
| Low | 1.25 (0.66–2.35) | 1.14 (0.60–2.15) | ||
| High | 0.46 (0.25–0.84) | 0.41 (0.22–0.76) | ||
| Ptrendb | 0.038 | 0.014 | ||
Abbreviation: CI, confidence interval; OR, odds ratio.
The ordinal logistic regression analysis model initially included age, sex, year of diagnosis, family history of colorectal carcinoma in parent or sibling, tumor location, tumor differentiation, microsatellite instability, CpG island methylator phenotype, KRAS, BRAF, and PIK3CA mutations, and LINE-1 methylation level. A backward stepwise elimination with a threshold of P < 0.05 was used to select variables in the final models. Variables remaining in the final multivariable ordinal logistic regression models are shown in eTable 3 in the Supplement.
Ptrend value was calculated by the linear trend across the ordinal (negative, low, and high) categories of the amount of Fusobacterium nucleatum as a continuous variable in the ordinal logistic regression model for the density of CD3+, CD8+, CD45RO+, or FOXP3+ T-cells (an ordinal quartile outcome variable). Because we assessed four primary outcome variables, we adjusted two-sided α level to 0.013 (= 0.05/4) by simple Bonferroni correction.