Outcome predictors in SARS-CoV-2 disease (COVID-19): The prominent role of IL-6 levels and an IL-6 gene polymorphism in a western Sicilian population.

Dear Editor

Recently in this Journal, Grifoni and colleagues reported that IL-6 levels at hospital admission seem to be the best prognosticator for negative outcomes in patients with coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Considering the preponderant role of host response in influencing the clinical evolution of SARS-CoV-2 infection, we studied polymorphisms of the IL-6 gene -174G/C (rs1800795), as well as its receptor IL-6R (rs2228145), TNF-α (rs1800629), ANGP2 (rs55633437), and MX1 (rs2071430) in patients hospitalized with COVID-19 to identify any genetic predisposition to a worse outcome. Other clinical features, such as IL-6 serum level, and risk factors for severity and mortality were also analyzed. We analyzed 316 Sicilian (Italy) patients, divided into two groups according to the type of oxygen and ventilation therapy they required during hospitalization: (1) non-critical (n = 271), patients admitted for reasons other than SARS-CoV-2 pneumonia with a positive preadmission screening, and patients hospitalized for SARS-CoV-2 infection without respiratory insufficiency; (2) critical (n = 45), subjects who underwent non-invasive ventilation (Suppl. Materials).

Table 1 shows some demographic features, the P/F ratio (pO/FiO) at admission, and the PSI as indicators of the severity of respiratory involvement in the two groups. Men were more prevalent in the two groups, but without statistical significance. Average age significantly increased with increasing disease severity, in line with the literature data. , , The P/F ratio, indicative of the degree of respiratory failure at admission, was confirmed to be related to disease outcome (p < 0.0001). The PSI, split into two severity groups, was found to reflect the degree of disease impairment, being significantly less severe at baseline (grade 2) in non-critical than in critical subjects (p < 0.0001). Instead, other indicators of the severity of pulmonary involvement, such as absence of pneumonia, pulmonary thickening, pulmonary consolidation, ground glass, and mixed CT features, did not show statistically significant differences between the groups (Table 1).

Table 1

Demographic characteristics, blood gas analysis parameters, presence of comorbidities, laboratory parameters and characteristics of chest CT scan images of the study population, divided by degree of respiratory involvement.

	Non-critical	Critical
	N = 271 (%)	N = 45 (%)	p <
Sex (M;%)	158 (58.3)	23 (51.1)	n.s.
Age (years; mean ± SD)	64.06 ± 14.1	71.1 ± 15.6	0.003
P/F	331.4 ± 90.1	261.9 + 106.6	0.0001
PSI
1 (96)	187 (73.6)	16 (41)
2 (%)	67 (26.4)	23 (59)	0.0001
COPD	15 (5.5)	1(2.2)	n.s.
Diabetes	58 (21.4)	7(15.6)	n.s.
Hypertension	154(65.8)	28 (62.2)	n.s.
Ischemic heart disease	41 (15)	17 (37)	0.001
IL-6 (pg/ml)	84.78 (1.5–1314)	1146(7.1–24526)	0.0001
D-dimer (ng/ml)	740 (82–64169)	867 (185–50386)	0.005
CRP (mg/l)	17.6 (0.5–307.26)	30.235 (0.88–201.52)	n.s.
NLR	4.529 (0.48–39.9)	6.33 (0.0008–22.5)	0.05
PCT (ng/l)	0.084 (0.02–40)	0.108 (0.03–55.26)	n.s.
Platelets (mm3)	224000 (31000–575000)	195000(44000–337000)	0.03
NT-pro-BNP (ng/l)	186 (10–19514)	337 (28.8–33519)	n.s.
WBC (mm3)	7590 (3040–46510)	7565 (1400–19610)	n.s.
Absence of pneumonia	29 (10.9)	3 (7.5)	n.s.
Pulmonary thickening	3(1)	1 (2.5)	n.s.
Pulmonary consolidation	1 (0.4)	2(5)	n.s.
Ground glass	5 (1.9)	3 (7.5)	n.s.
Mixed CT features	229 (85.8)	31 (77.5)	n.s.

P/F = pO2/FiO2; PSI = Pneumonia Severity Index; COPD = Chronic obstructive pulmonary disease; IL-6 = nterleukin 6; CRP = C-reactive protein; NLR = Neutrophil-Lymphocyte Ratio; PCT = Procalcitonin; NT-pro-BNP = N-terminal pro-hormone of brain natriuretic peptide; WBC = white blood cells

The presence of comorbidities such as COPD, diabetes, and hypertension did not reach statistically significant percentages in the two groups (Table 1). Only the presence of ischemic heart disease was found to be significantly correlated with disease severity(p < 0.001) (Table 1), a finding already described in the literature, with cardiovascular disease being one of the most important underlying diseases affecting the prognosis of patients with COVID-19.

In addition, we found significantly lower platelet counts in subjects with more severe disease (p < 0.03). Furthermore, other parameters known to be related to the worst outcome of the disease (i.e., D-dimer and NLR) , were confirmed in our series to be significantly related (p < 0.005 and p < 0.05, respectively) (Table 1). Finally, in accordance with Grifoni et al., we found maximum serum IL-6 levels, which were significantly higher in critically ill than non-critical subjects (p < 0.0001) (Table 1), to be strongly correlated with the worst prognosis among the markers of inflammatory status.

Successively, according to the literature data cut-off value of serum IL-6 > 55 pg/mL for identifying patients at high risk of severe COVID 19 forms, we observed higher IL-6 values in 17/105 (16%) non-critical and 18/33 (53%) critical patients (OR 5.8 CI 95% 2.4-13.6).

For the genetic study, IL-6 rs1800795, IL-6R rs2228145, TNF-α rs1800629, ANGP2 rs55633437, and MX1 rs2071430 genotype distribution in critical and non-critical subjects is shown in Table 2 and Suppl. Table 1, and all followed the Hardy-Weinberg equilibrium. The CC genotype frequency for IL-6 rs1800795 was significantly more frequent in critically ill than non-critical patients (p = 0.044).A significantly greater association was also observed between the rs1800795 polymorphism in critical subjects in the recessive model (CC vs. GG+CG) than in non-critical patients (p = 0.044) (Table 2). No significant associations were found in genotype frequencies for the other 4 SNPs analyzed (Suppl. Table 1).

Table 2

Genotype distribution of IL-6 variants in the non-critical and critical groups.

		Non-critical	Critical
	IL-6 - rsl800795	N = 103 (%)	N = 33 (%)	p
	GG	69 (67.0)	20 (60.6)	n.s.
Codominant model	CG	33 (32.0)	10 (39.3)	n.s.
	CC	1 (1.0)	3(9.1)	0.044
Dominant model	GG CG+CC	69 (67.0) 34 (33.0)	20 (60.6) 13 (39.4)	n.s.
Recessive model	GG+CG CC	102 (99.0) 1 (1.0)	30 (90.9) 3(9.1)	0.044

Since the rs1800795 variant of IL-6 affects gene transcription and serum IL-6 levels, we analyzed the association between this variant's presence and IL-6 serum levels (N = 138). In accordance with Smieszek et al., no significant correlation between IL-6 levels and the IL-6 rs1800795 genotype was found (Suppl. Fig. 1A). In addition, since some studies have reported that IL-6R variants may correlate with circulating IL-6 levels, we correlated the rs2228145 genotype with IL-6 serum levels, finding no significant correlation between IL-6 levels and the IL-6R rs2228145 genotype (Suppl. Fig. 1B).

Subsequently, we examined the cumulative effects of the selected SNPs, developing a genetic risk score (GRS) by summing the number of risk alleles. For each SNP, a score of 0 was defined for homozygous non-risk alleles, 1 for heterozygous risk and non-risk alleles, and 2 for two homozygous risk alleles. A higher mean risk score was significantly associated with critical patients when the sum of the five scores for the rs1800629, rs1800795, rs2228145, rs2071430, and rs55633437 variants was considered for each patient (p = 0.026) (Suppl. Table 2). The mean of the gene count score was 3.67 ± 1.26 in the non-critical group and 4.26 ± 1.13 in the critical group. The cumulative effects of 4, 3, and 2 variants were also analyzed (Suppl. Table 2). After removing genetic variants, the contribution of IL-6 rs1800795 appeared to be crucial for risk prediction; in all cases where critical patients showed significant differences in GRS compared to non-critical patients, the IL-6 gene variant was present. Moreover, in all cases of significant differences between critical and non-critical patients, the MX1 variant was always present together with the IL-6 variant, except in one case (i.e., for the 4 variants IL-6, MX1, ANGP2, and TNF-α) in which no significant difference was observed (p = 0.0516), possibly due to the small number of patients. This suggests that the copresence of the MX1 and IL-6 variants could confer greater risk of disease severity.

Our study confirmed the prominent role of IL-6 levels and the genetic predisposition of an IL-6 gene variant in response to SARS-CoV-2 infection as predictors of COVID-19 disease severity and unfavorable outcomes, as well as the role of age and ischemic heart disease as important negative prognostic factors.