Table 3.
Stepwise multiple regression
| Stepwise multiple regression | Predictor variables | ||||||
|---|---|---|---|---|---|---|---|
| |
Model |
|
|
|
Variable |
Beta |
P |
|
F |
df |
P |
R2 |
|
|
|
|
|
B cells all |
21.650 |
3 |
0.000 |
0.391 |
CXCL13 |
0.745 |
0.000 |
| CXCL12 |
−0.490 |
0.001 |
|||||
| BAFF |
0.320 |
0.012 |
|||||
|
B cells |
18.098 |
2 |
0.000 |
0.262 |
CXCL13 |
0.616 |
0.000 |
| |
|
|
|
|
CXCL12 |
−0.271 |
0.010 |
|
Plasmablasts |
20.265 |
2 |
0.000 |
0.287 |
CXCL13 |
0.413 |
0.000 |
| |
|
|
|
|
BAFF |
0.234 |
0.009 |
|
T cells |
6.001 |
2 |
0.003 |
0.110 |
BAFF |
−0.498 |
0.001 |
| |
|
|
|
|
APRIL |
0.296 |
0.05 |
| Monocytes | No variables included | ||||||
Stepwise multiple regression was applied to determine the most predictable cytokine/ chemokine for the presence of certain immune cells in the CSF. Variables (cytokines/ chemokines), which fulfilled the entry criteria of P <0.05 for the regression model, are displayed. For B cells and plasmablasts, CXCL13 turned out to be the most predictable chemokine.