Table 3.
Immunological correlates of VMB clusters
| Correlates | KRST-I (n = 19) | KRST-II (n = 136) | KRST-pIII-V (n = 53) | p value *a | |||||
|---|---|---|---|---|---|---|---|---|---|
| (in log10) | Mean (95% CI) | OR/aOR (95% CI) | Mean (95% CI) | OR (95% CI) | aOR (95% CI) | Mean (95% CI) | OR (95% CI) | aOR (95% CI) | |
| IL-1α | 1.08 (0.86, 1.3) | Ref | 1.22 (1.14, 1.31) | 1.61 (0.69, 3.77) | 1.58 (0.58, 4.33) | 1.64 (1.48, 1.81) | 7.4 (2.63, 20.82) | 11.5† (3.29, 40.22) | <0.0001 |
| IL-1β | 0.65 (0.33, 0.97) | Ref | 0.83 (0.71, 0.95) | 1.4 (0.72, 2.74) | - | 1.34 (1.11, 1.57) | 3.33 (1.58, 7.03) | - | <0.0001 |
| IL-1rab | 4.65 (4.36, 4.94) | Ref | 4.83 (4.74, 4.91) | 1.9 (0.76, 4.76) | - | 4.87 (4.74, 5) | 2.28 (0.79, 6.6) | - | 0.28 |
| IL-6 | 0.64 (0.33, 0.95) | Ref | 0.9 (0.79, 1) | 1.68 (0.85, 3.3) | - | 0.9 (0.68, 1.11) | 1.68 (0.8, 3.54) | - | 0.31 |
| IL-8 | 2.06 (1.83, 2.28) | Ref | 2.19 (2.09, 2.28) | 1.41 (0.62, 3.2) | - | 2.25 (2.04, 2.45) | 1.65 (0.68, 3.97) | - | 0.54 |
| IL-12c | 0.07 (−0.18, 0.32) | Ref | 0.13 (0.05, 0.21) | 1.31 (0.45, 3.81) | - | 0.24 (0.12, 0.35) | 2.11 (0.66, 6.7) | - | 0.28 |
| IP-10 | 2.54 (2.36, 2.73) | Ref | 2.62 (2.51, 2.73) | 1.22 (0.57, 2.6) | 0.98 (0.39, 2.43) | 2.15 (1.93, 2.38) | 0.47 (0.21, 1.06) | 0.16† (0.05, 0.47) | 0.0002 |
| GM-CSF | 0.26 (0.05, 0.47) | Ref | 0.32 (0.25, 0.39) | 1.38 (0.47, 4.03) | - | 0.13 (−0.01, 0.27) | 0.58 (0.19, 1.78) | - | 0.0351 |
| G-CSF | 1.84 (1.57, 2.11) | Ref | 1.91 (1.79, 2.02) | 1.16 (0.57, 2.38) | 0.71 (0.26, 1.92) | 2.08 (1.93, 2.23) | 1.83 (0.81, 4.15) | 3.6†† (1.02, 12.73) | 0.18 |
| MIP-1βd | 0.69 (0.23, 1.15) | Ref | 0.94 (0.83, 1.05) | 1.57 (0.84, 2.94) | 1.73 (0.79, 3.82) | 0.7 (0.48, 0.93) | 1.02 (0.52, 1.98) | 0.61 (0.25, 1.52) | 0.085 |
Abbreviations: CI confidence interval, KRST Kenya, Rwanda, South Africa, Tanzania, OR odds ratio, aOR adjusted odds ratio. For cytokine and growth factor abbreviations, we refer to the text.
*One way analysis of variance test in a bivariable analysis.
aSix variables qualified for inclusion in the multivariable model but only four (IL-1α, IP-10, G-CSF, and MIP-1β) variables remained in the final model after the model selection process. Although, IL-1β was highly significant in the bivariable analysis, it was eliminated during the model selection because it was highly correlated with IL-1α (Pearson’s correlation coefficient = 0.65).
b19/208 values missing.
cStronger evidence of association was observed when pregnant women and Tanzanian women (stage of menstrual cycle at time of sampling not known) were excluded (p = 0.09).
dWeaker evidence of association was observed when pregnant women and Tanzanian women (stage of menstrual cycle at time of sampling not known) were excluded (p = 0.27).
†Adjusted P-value < 0.001.
††Adjusted P-value = 0.01.