Table 4 Performance of each candidate markers dichotomized into lower than or higher than the cutpoint or of the combination of HGF and CXCL13 to predict death during the follow-up of COVID-19 patients enrolled in LUH-1, LUH-2 and the FCS cohorts.

From: The cytokines HGF and CXCL13 predict the severity and the mortality in COVID-19 patients

Marker

Low

High

p-value‡

Hazard ratio*

p-value||

HGF

5 (4.6)

13 (14.9)

0.012

1.53 (0.29–8.18)

0.621

CXCL13

2 (2.4)

16 (14.0)

0.005

4.94 (0.85–28.6)

0.075

CXCL9

5 (4.6)

13 (14.6)

0.016

1.02 (0.32–3.26)

0.980

IL-6

10 (7.1)

8 (14.3)

0.114

1.33 (0.45–3.87)

0.606

CCL2

12 (8.1)

6 (12.5)

0.352

0.66 (0.21–2.03)

0.463

CXCL10

9 (6.7)

9 (14.5)

0.076

3.73 (1.14–12.2)

0.029

IL-1RA

8 (6.3)

10 (14.3)

0.063

2.39 (0.73–7.82)

0.151

CCL4

2 (4.6)

16 (10.5)

0.230

2.57 (0.48–13.7)

0.269

VEGF-A

8 (8.0)

10 (10.3)

0.574

1.23 (0.40–3.74)

0.721

IL-15

11 (8.7)

7 (9.9)

0.792

0.85 (0.28–2.58)

0.780

IL-10

13 (8.5)

5 (11.4)

0.561

0.81 (0.26–2.50)

0.712

IL-1β

12 (10.1)

6 (7.7)

0.569

0.45 (0.15–1.36)

0.158

LIF

12 (8.1)

6 (12.2)

0.384

0.74 (0.24–2.26)

0.597

Combination of HGF and CXCL13

HGF/CXCL13

1 (1.5)

17 (13.3)

0.006

8.80 (0.96–80.3)

0.054

  1. The first two columns indicate the percentage of subjects within a given category (low or high levels) who died during follow-up, all cohorts together.
  2. *Adjusted for age (continuous), ICU stay (yes/no) and cohort (Lausanne 1/Lausanne 2/Paris), ‡analysis by chi-square; ‖, analysis by a multilevel survival model using a Weibull distribution, where patients were nested within each cohort.
Back to article page