Interleukin-1 Receptor Antagonist Gene (IL1RN) Variants Modulate the Cytokine Release Syndrome and Mortality of COVID-19

Abstract

Background

We examined effects of single-nucleotide variants (SNVs) of IL1RN, the gene encoding the anti-inflammatory interleukin 1 receptor antagonist (IL-1Ra), on the cytokine release syndrome (CRS) and mortality in patients with acute severe respiratory syndrome coronavirus 2 (SARS-CoV-2) infection.

Methods

IL1RN CTA haplotypes formed from 3 SNVs (rs419598, rs315952, rs9005) and the individual SNVs were assessed for association with laboratory markers of inflammation and mortality. We studied 2589 patients hospitalized with SARS-CoV-2 between March 2020 and March 2021.

Results

Mortality was 15.3% and lower in women than men (13.1% vs 17.3%, P = .0003). Carriers of the CTA-1/2 IL1RN haplotypes exhibited decreased inflammatory markers and increased plasma IL-1Ra. Evaluation of the individual SNVs of the IL1RN, carriers of the rs419598 C/C SNV exhibited significantly reduced inflammatory biomarker levels and numerically lower mortality compared to the C/T-T/T genotype (10.0% vs 17.8%, P = .052) in men, with the most pronounced association observed in male patients ≤74 years old, whose mortality was reduced by 80% (3.1% vs 14.0%, P = .030).

Conclusions

The IL1RN haplotype CTA and C/C variant of rs419598 are associated with attenuation of the CRS and decreased mortality in men with acute SARS-CoV-2 infection. The data suggest that the IL1RN pathway modulates the severity of coronavirus disease 2019 (COVID-19) via endogenous anti-inflammatory mechanisms.

Graphical Abstract

SARS-CoV-2, inflammation, IL1RN, genetic variation and mortality

Severe coronavirus disease 2019 (COVID-19) resulting from acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection may cause acute respiratory distress syndrome (ARDS), multiorgan failure, and death in nearly 10%–20% of cases [1, 2]. Patients with these COVID-19 symptoms exhibit marked elevations of plasma cytokine levels similar to those observed in the cytokine release syndrome (CRS), including, though not limited to, elevations of interleukin 1β (IL-1β), IL-2, and IL-6 [3–5]. Various factors, including infections, immunotherapy drugs, and cell therapy, can trigger CRS. Infections can stimulate an aggressive immune response, leading to CRS, while immunotherapy drugs and cell therapy can also induce CRS as a side effect of the treatment [6].

We have shown that specific IL1RN haplotypes composed of the 3 variants rs419598, rs315952, and rs9005 are associated with inflammation and disease severity in osteoarthritis and rheumatoid arthritis [7]. Notably, rheumatoid arthritis patients carrying the IL1RN CTA haplotype exhibit lower disease activity scores, plasma C-reactive protein (CRP), and IL-6 associated with increased IL-1Ra [7]. We sought to determine whether these IL1RN genetic variants (single-nucleotide variants [SNVs] rs419598, rs315952, and rs9005) are similarly associated with decreased hyperinflammation and mortality in patients with moderate to severe SARS-CoV-2. We performed low-coverage whole-genome sequencing followed by imputation on 2589 patients admitted to New York University (NYU) Langone's Tisch Hospital from 1 March 2020 through 1 March 2021. We analyzed the association of these IL1RN genotypes with inflammatory markers of the CRS and mortality.

METHODS

Patient Population

This is a retrospective, observational cohort of adult patients (aged 19 years and older) admitted to Tisch Hospital, 1 of 3 hospitals within the NYU Langone Health system. Age, sex, and body mass index (BMI)-matched healthy controls for multiplex cytokines screening were recruited from our osteoarthritis clinics; these individuals did not have inflammatory comorbidities and were never exposed to SARS-CoV-2 (IRB protocol No. i05-131). The Institutional Review Board at NYU Grossman School of Medicine approved the protocol (i05-131, S16-00122 and i20-00485).

Study Cohort

We identified patients tested for SARS-CoV-2 and positive for real-time reverse transcriptase-polymerase chain reaction (RT-PCR) assays on nasopharyngeal or oropharyngeal swab specimens. Our clinical laboratory conducted tests using the Roche SARS-CoV-2 assay in the Cobas 6800 instruments through emergency use authorization granted by the US Food and Drug Administration (FDA). On 31 March 2020, we added testing using the SARS-CoV-2 Xpert Xpress assay in the Cepheid GeneXpert instruments under emergency use authorization by the FDA. The targets amplified by these assays are the ORF1/a and E genes in the Roche Cobas assay and N2 and E genes in the Cepheid Xpert Xpress. (Additional data source information are available in Supplemental Methods).

Sample Processing, Genotyping, and Imputation

Discarded blood samples for clinical use were utilized in this study. Total genomic DNA was isolated using standard protocols and used to generate 1.2× low-coverage human whole-genome sequences using standard protocols [8] in the NYU Langone Genome Technology Center. After quality control, these data were used to impute all common (minor allele frequency ≥1%) SNV genotypes for each sample, using a reference population of 7345 samples from 26 distinct geographic populations of the world (https://gencove.org); these analyses were conducted by Gencove (New York, NY) and have been demonstrated to produce genotype data with nonreference allele concordance ≥98% [8]. We extracted IL1RN (rs419598, rs315952, and rs9005) genotypes for this study; allele frequencies were estimated using standard allele counting. Plasma cytokines IL-1β, IL-2, and IL-6 were determined by a test developed by ARUP Laboratories (Salt Lake City, UT) and approved by the New York State Department of Health. Multiplex assay and haplotype information are provided in Supplementary Methods.

Primary Outcomes and Study Variables

Patient outcomes of interest included inflammatory markers for evidence of CRS and mortality as evidence of severe coronavirus disease 2019 (COVID-19) outcomes. Other study variables included age, sex, self-identified race and ethnicity, BMI, and preexistent comorbidities.

Statistical Analysis

We assessed demographic variables, laboratory values, and mortality stratified by IL1RN genotypes/haplotypes, overall and separately for men and women. Continuous variables were summarized using means and standard deviations, whereas categorical variables were summarized using frequency and proportions. Summaries were created to assess demographic variables and laboratory values in the overall study sample and by mortality status (alive and deceased), by sex (men and women), and by race/ethnicity. Univariate parametric tests (a t test for continuous variables and c² tests for categorical variables) were used to assess CRS and mortality outcomes by each stratum. Separate age, sex, and BMI-adjusted logistic regressions were fitted to estimate the odds ratios (ORs) and 95% confidence interval (CI) for predicting mortality with each genotype/haplotype as the primary covariate. Significance tests were 2-tailed at a significance level of .05. For multiple comparisons, P values from univariate tests were adjusted at a 5% false discovery rate using the Benjamin-Hochberg criterion, separately for demographics and CRS category.

To compare in-hospital mortality between IL1RN genotypes, we used multivariable logistic regression, adjusting for age, sex, and BMI in a stepwise procedure. Age was adjusted first as a continuous variable. To account for nonlinear relationships between age and COVID-19 mortality, we adjusted for categorical age in deciles as a secondary analysis. ORs and 95% CIs are reported. Analyses were repeated separately for sex-stratified samples to assess differences in relationships between men and women, as well as by age (≥74 and ≥75 years) to assess differences in associations by age. The statistical software R was used for the statistical analysis.

RESULTS

Patient Population

We obtained laboratory and clinical information from 2589 hospitalized COVID-19 PCR-positive patients admitted to NYU Langone's Tisch Hospital between 1 March 2020 and 1 March 2021. The Alpha and Beta SARS-CoV-2 variants were the most prevalent during this time. Low-coverage (1.2×) whole-genome sequencing data followed by imputation were obtained on all 2589 patients, from which IL1RN genotype data were obtained, as described elsewhere [8]. The mean age and BMI for the cohort were 61.2 years (SD 18.66; range, 19–102 years) and 30.43 (SD 7.71), respectively. Men represented 53.3% of the sample. IL1RN rs419598, rs315952, and rs9005 genotype data were available for all patients. Biomarkers noted in the clinical electronic hospital record (EHR) for IL-1β, IL-2, and IL-6 were available for 642, 645, and 1229 subjects, respectively, whereas other plasma inflammatory markers were available for more than 2000 subjects (Supplementary Table 1).

Inflammatory Markers Associated With Severe SARS-CoV-2 Infection

Hospitalized patients with SARS-CoV-2 exhibited marked elevated levels of multiple cytokines (IL-1β, IL-2, IL-1Ra, and IL-6) and inflammatory markers (CRP, D-dimer, procalcitonin, and ferritin), as previously reported [9, 10]. Because multiple cytokines of interest were not measured in the course of routine patient care reported in the EHR, we analyzed plasma from 359 patients (all samples from NYU Center for Biospecimen Research and Development biobank) from the 2589-patient cohort using a multiplex enzyme-linked immunosorbent assay (ELISA) assay, and compared to age, sex, and BMI matched healthy controls (n = 22). The results showed marked elevations of additional cytokines not available in the EHR, including IL-1α, IL-5, IL-8, IL-17, tumor necrosis factor-α (TNF-α), vascular endothelial growth factor (VEGF), and interferon-α. Elevations of IL-6, IL-1Ra, IL-8, and IL-10 exceeded 10-fold normal values based on our controls (Supplementary Table 2) and reported in the literature [5, 11–13].

There were 397 (15.3%) deaths among the 2589 patients. As expected, age, male sex (Figure 1), and BMI (Table 1) were associated with increased mortality. CRS-associated inflammatory biomarkers were elevated in both patients who survived and died; however, deceased patients had significantly higher levels of IL-6, CRP, procalcitonin, ferritin, and D-dimer, as well as reduced levels of complement components C3 and C4 (Table 1).

Figure 1.

IL1RN rs419598 protective genotype is associated with decreased mortality among men. A, Mortality rate by age categories and sex among SARS-CoV-2 infected COVID-19 patients. B, Model predicted mortality rate for men by IL1RN rs419598 genotype. Unadjusted ORs and 95% CIs were estimated using logistic regression adjusted for age or age and BMI. C, Mortality rate among men by IL1RN rs419598 genotype (C/C vs C/T-T/T) in different age groups. Abbreviations: BMI, body mass index; CI, confidence interval; COVID-19, coronavirus disease 2019; IL1RN, interleukin-1 receptor antagonist gene; OR, odds ratio; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2. P-value of overall mortality among age groups (age 19-44 as the reference group) **P<0.001.

Table 1.

Inflammatory Biomarkers Associated With COVID-19 Mortality (N = 2589)

	Deceased (n = 397), Mean (SD)	Alive (n = 2192), Mean (SD)	FDR P Value
Demography
Age, y	74.08 (12.71)	58.87 (18.61)	.0002
BMI	29.50 (7.11)	30.60 (7.80)	.0060
Sex, Male,  No. (%)	240 (60.5)	1141 (52.1)	.0032
Inflammation markers
Lab IL-1Ra	2870 (2350)	2900 (2450)	.9720
Lab IL-1β	8.71 (16.34)	6.37 (6.60)	.0986
IL-1β, max	8.38 (16.19)	7.05 (13.07)	.4369
Lab IL-2	7.95 (13.22)	6.68 (22.65)	.6983
IL-2, max	8.03 (15.61)	7.38 (40.46)	.9720
Lab IL-6	72.23 (235.62)	29.02 (85.41)	.0114
IL-6, max	212.31 (801.29)	65.51 (199.82)	.0005
Lab CRP	155.98 (94.22)	100.14 (82.81)	.0003
CRP, max	244.15 (109.33)	130.57 (99.60)	.0003
Lab procalcitonin	6.26 (41.79)	0.81 (7.36)	.0409
Lab ferritin	1431.22 (2353.44)	1134.66 (2245.36)	.0258
Lab D-dimer	2580.17 (5944.84)	1066.20 (2993.75)	.0003
D-dimer, max	8174.42 (11698.82)	2234.18 (5472.01)	.0003
Lab C3 complement	102.34 (34.71)	130.67 (39.06)	.0003
Lab C4 complement	27.04 (14.72)	32.63 (14.77)	.1018
Lab troponin	0.23 (0.85)	0.18 (3.11)	.6983

Summary of demographics, inflammation, and clinical biomarkers among deceased and alive patients. The BMI is the weight in kilograms divided by the square of the height in meters. Biomarkers are presented as mean (SD). FDR P values control for FDR at 5% using Benjamin-Hochberg criteria. Lab, routine laboratory tests on admission; Max, maximum values of inflammation markers during hospitalization.

Abbreviations: BMI, body mass index; CRP, C reactive protein; FDR, false discovery rate; HDL, high density lipoprotein; IL, interleukin.

Inflammatory Markers Are Decreased in Carriers of the IL1RN CTA Haplotype

We compared evidence of hyperinflammation in carriers of IL1RN CTA-1/2 haplotype (either or two copies of CTA haplotype) CTA. CTA and non-CTA that can be constructed from these 3 SNVs with a frequency of >5% and are found on the same locus. The carriers of the CTA-1/2 haplotypes exhibited significantly lower levels of inflammatory markers (IL-1β, IL-2, IL-6, D-dimer) and higher levels of IL-1Ra relative to non-CTA haplotype carriers (Table 2). The higher levels of the anti-inflammatory protein IL-1Ra, associated with lower levels of IL-6 in CTA carriers, are consistent with our prior observations in patients with rheumatoid arthritis [7]. While mortality was reduced by 40% in men with CTA haplotypes in the ≤74 year age groups (13.6% vs 8.1%), the age and BMI adjusted association was not significant (OR = 0.58; 95% CI = .28–1.09; P = .11; Table 2).

Table 2.

Patients With the IL1RN CTA Haplotype Exhibit Lower Inflammatory Biomarker Levels Among SARS-CoV2–Infected Patients

	All Age Groups (n = 2181)			Age ≤74 y (n = 1622)			Age ≥75 y (n = 558)
	CTA-1/2 (n = 263)	CTA-0 (n = 1918)		CTA-1/2 (n = 205)	CTA-0 (n = 1417)		CTA-1/2 (n = 58)	CTA-0 (n = 500)
	Mean (SD)	Mean (SD)	FDR P Value	Mean (SD)	Mean (SD)	FDR P Value	Mean (SD)	Mean (SD)	FDR P Value
Demography
Age, y	58.44 (18.25)	61.37 (18.58)	.0453	51.58 (14.30)	53.64 (15.04)	.1686	82.66 (5.51)	83.31 (5.63)	.6041
BMI	30.48 (6.82)	30.44 (7.48)	.9193	30.99 (6.63)	35.73 (161.73)	.2768	28.74 (7.10)	27.62 (6.04)	.6041
Sex, male, No. (%)	150 (57.0)	1008 (52.6)	.2955	124 (60.5)	777 (54.9)	.2217	27 (45.8)	230 (46.3)	.9527
Inflammation markers
Lab IL-1Ra	3910 (3630)	2700 (2040)	.0458	3415 (3514)	2535 (1900)	.1737	6301 (3420)	3138 (2337)	.1177
Lab IL-1β	5.85 (3.02)	7.32 (11.43)	.0436	5.78 (3.05)	7.42 (11.02)	.0919	6.13 (2.88)	7.04 (12.54)	.6202
Lab IL-2	5.10 (2.28)	7.47 (23.03)	.0458	5.10 (2.18)	7.68 (25.47)	.1412	5.09 (2.58)	6.86 (13.51)	.3426
Lab IL-6	27.56 (55.97)	42.91 (179.57)	.0623	27.47 (59.02)	43.53 (183.16)	.1412	29.20 (48.99)	41.34 (170.22)	.5476
IL-1β, max	5.89 (2.72)	7.89 (15.74)	.0302	5.75 (2.67)	8.12 (16.67)	.0884	6.41 (2.83)	7.23 (12.62)	.6202
IL-2, max	4.63 (2.39)	7.98 (39.71)	.0828	4.66 (2.25)	8.44 (45.30)	.1736	4.54 (2.83)	6.65 (13.46)	.3155
IL-6, max	84.24 (276.51)	111.86 (455.03)	.3443	85.69 (280.78)	122.31 (516.41)	.3031	74.88 (142.39)	85.60 (237.80)	.7195
Lab CRP	103.79 (75.66)	111.66 (88.14)	.1918	105.77 (79.48)	115.72 (90.59)	.1935	96.51 (61.52)	101.85 (80.85)	.6202
CRP, max	136.20 (99.39)	152.72 (110.11)	.0458	140.48 (105.02)	156.99 (113.09)	.1412	126.95 (80.26)	144.09 (101.62)	.3274
Lab procalcitonin	0.45 (0.91)	1.98 (20.10)	.0186	0.74 (3.62)	2.47 (23.92)	.1060	0.44 (0.91)	0.86 (3.68)	.1503
Lab ferritin	961.24 (1189.92)	1272.14 (2662.63)	.0186	1138.48 (1711.25)	1347.06 (2870.61)	.2244	721.59 (928.99)	1102.34 (2099.74)	.1177
Lab D-dimer	933.97 (1825.80)	1367.67 (3814.58)	.0237	851.04 (1659.62)	1154.14 (3368.53)	.1412	999.41 (1859.41)	1859.46 (4666.31)	.1177
D-dimer, max	2052.95 (3430.96)	3476.49 (7614.07)	.0013	2117.34 (3750.73)	3473.50 (7931.16)	.0065	2134.29 (3794.72)	3489.35 (6867.39)	.1177
Mortality, No. (%)a	31 (11.8)	298 (15.5)	0.83 (.54–1.26); .401	14 (6.8)	149 (10.5)	0.62 (.33–1.05); .097	17 (28.8)	149 (30.0)	0.94 (.50–1.71); .852
Mortality, men only, No. (%)a	17 (11.3)	180 (17.9)	0.73 (.41–1.24); .271	10 (8.1)	106 (13.6)	0.58 (.28–1.09); .114	7 (25.9)	75 (32.7)	0.75 (.28–1.80); .543

Summary of demographics and inflammation biomarkers by haplotypes. The BMI is the weight in kilograms divided by the square of the height in meters. Biomarkers are presented as mean (SD). FDR P values control for false discovery rate at 5% using Benjamin-Hochberg criteria within each subgroup of biomarkers.

Abbreviations: BMI, body mass index; CRP, C-reactive protein; FDR, false discovery rate; IL, interleukin; Lab, routine laboratory tests on admission; Max, maximum values of inflammation markers during hospitalization.

^aFor mortality, comparison between haplotype odds ratio (adjusted odds ratio [95% confidence interval]; P value) derived from multivariable logistic regression adjusting for age, sex, and BMI; and for age and BMI for sex-specific mortality.

Inflammatory Markers Are Decreased in Carriers of the IL1RN rs419598 C/C SNV

We next individually evaluated each SNV of the IL1RN CTA haplotype (rs419598, rs315952, and rs9005). As shown in Table 3, multiple biomarkers of inflammation in COVID-19 were significantly lower in the patients with the IL1RN rs419598 C/C SNV compared to those with C/T or T/T. Conversely, like the CTA haplotype findings, and in contrast to decreased inflammatory cytokine levels observed in IL1RN rs419598 C/C individuals, these patients exhibited numerically higher plasma IL-1Ra levels, the gene product of IL1RN and an endogenous anti-inflammatory cytokine (Table 3). We next compared the effects of the IL1RN rs419598 C/C genotype on plasma inflammatory biomarkers levels and mortality separately in men and women (Tables 3). The IL1RN rs419598 C/C versus C/T-T/T genotype-dependent decrease of inflammation-associated biomarkers did not differ in both men and women, except for maximum IL-6- and CRP, for which no significant differences between genotypes were observed in women (Table 3). Table 3 shows that the IL1RN rs419598 C/C SNV was numerically associated with a decreased trend in mortality in men (10.0% vs 17.8%; OR = 0.49; 95% CI = .23–1.00; P = .052), but not in women.

Table 3.

Men with the IL1RN rs419598 C/C Genotype Exhibited Lower Inflammation Cytokine Levels and Mortality

	IL1RN rs419598			Men (n = 1380)			Women (n = 1208)
	C/C (n = 175)	C/T-T/T (n = 2414)		C/C (n = 90)	C/T-T/T (n = 1290)		C/C (n = 85)	C/T-T/T (n = 1123)
	Mean (SD)	Mean (SD)	FDR P Value	Mean (SD)	Mean (SD)	FDR P Value	Mean (SD)	Mean (SD)	FDR P Value
Demography
Age, y	60.5 (19.5)	61.3 (18.6)	.619	62.0 (17.7)	62.5 (15.9)	.8190	58.9 (21.3)	59.8 (21.2)	.6810
BMI	30.2 (6.84)	30.5 (7.77)	.659	29.0 (5.56)	30.0 (6.67)	.1760	31.6 (7.80)	30.9 (8.84)	.6810
Sex, male No. ( %)	90 (51.4%)	1290 (53.4%)	.659
Inflammation markers
Lab IL-1Ra	3680 (3720)	2840 (2320)	.199	2410 (1600)	2930 (2340)	.2280	4950 (4750)	2730 (2290)	.1069
Lab IL-1β	5.55 (1.93)	7.13 (10.4)	.002	5.64 (2.32)	6.59 (8.45)	.1501	5.40 (0.949)	8.05 (13.0)	.0260
Lab IL-2	5.04 (1.92)	7.06 (20.6)	.024	5.16 (1.88)	7.18 (24.0)	.1501	4.83 (2.01)	6.85 (12.8)	.0854
Lab IL-6	24.1 (35.7)	41.7 (166)	.006	25.9 (41.4)	43.5 (168)	.0748	21.6 (25.9)	38.8 (164)	.0910
IL-1β, max	5.45 (0.853)	7.57 (14.1)	< .001	5.37 (0.765)	7.30 (14.4)	.0238	5.59 (1.00)	8.04 (13.6)	.0476
IL-2, max	4.58 (1.90)	7.51 (35.6)	.048	4.64 (1.74)	7.78 (43.2)	.1744	4.47 (2.21)	7.04 (14.8)	.0628
IL-6, max	58.9 (178)	112 (435)	.030	46.8 (84.1)	136 (529)	.0003	74.9 (256)	74.0 (215)	.9830
Lab CRP	96.4 (67.9)	111 (88.1)	.026	98.6 (65.3)	118 (89.6)	.0316	93.5 (71.8)	99.4 (84.6)	.7319
CRP, max	132 (95.4)	152 (111)	.026	133 (88.8)	162 (111)	.0234	129 (105)	136 (109)	.7800
Lab procalcitonin	0.349 (0.862)	1.74 (18.0)	.001	0.244 (0.525)	1.07 (6.11)	.0003	0.490 (1.16)	2.75 (27.5)	.0628
Lab ferritin	897 (1110)	1250 (2630)	.002	1240 (1330)	1500 (2730)	.1500	442 (374)	871 (2450)	.0013
Lab D-dimer	896 (1780)	1320 (3730)	.015	715 (1380)	1410 (4320)	.0003	1140 (2190)	1200 (2670)	.9078
D-dimer, max	1910 (3230)	3380 (7460)	< .001	1970 (3590)	3760 (7890)	.0003	1830 (2690)	2820 (6770)	.0628
Mortality, No. (%)a	20 (11.4)	377 (15.6)	.66 (.39–1.09); 0.12	9 (10.0)	230 (17.8)	.49 (.23–1.00); 0.052	11 (12.9)	147 (13.1)	1.0 (.48–2.09); 0.99

Summary of demographics and inflammation biomarkers by genotype. The BMI is the weight in kilograms divided by the square of the height in meters. Biomarkers are presented as mean (SD). FDR P values control for FDR at 5% using Benjamin Hochberg criteria within each subgroup of biomarkers.

Abbreviations: BMI, body mass index; CRP, C-reactive protein; FDR, false discovery rate; IL, interleukin; Lab, routine laboratory tests on admission; Max, maximum values of inflammation markers during hospitalization.

^aFor mortality, comparison between genotype odds ratio (adjusted odds ratio [95% confidence interval]; P value) derived from multivariable logistic regression adjusting for age, sex, and BMI; and for age and BMI for sex-specific mortality.

We also examined the association of IL1RN SNVs (rs315952, rs9005), part of the IL1RN CTA haplotype of interest [7]. Patients with the IL1RN rs315952 C/C genotype, reported by Meyer et al to reduce the risk of ARDS in critically ill patients [14], exhibited lower IL-1β and IL-2 (Supplementary Table 3) but did not affect other inflammatory markers. The SNV rs9005 A/A genotypes, in close linkage disequilibrium to rs419598 (D′/r2 0.81/0.61), were associated with elevations of IL-1Ra and decreased levels of IL-1β, IL-2, and D-dimer Max (Supplementary Table 4). Neither SNV was associated with decreased mortality.

The IL1RN C/C rs419598 SNV Is Associated With Significantly Decreased SARS-CoV-2 Mortality in Men Under the Age of 75 Years

As noted in Table 3, the C/C genotype was marginally associated with decreased male mortality and with lower levels of inflammatory biomarkers. As shown in Supplementary Table 5, female patients were younger and had higher BMI. Baseline and maximal elevations of IL-1β, IL-2, and IL-1Ra did not differ between men and women, while maximum IL-6 was lower in women compared to men, as has been reported [5] (74.04 vs 130.06, P = .026; Supplementary Table 5). Plasma levels of the inflammatory markers ferritin, CRP, and maximum D-dimer were also lower in women than men. Of note, mean plasma IL-1Ra levels exceeded 4950 pg/mL (Table 3), 20-fold higher than normal levels and 2-fold higher in women carriers of the C/C compared to carriers with the C/T-T/T genotype.

As shown in Figure 1A, the SARS-CoV-2 mortality rate increased with age and was higher in men than in women (OR, 1.55; 95% CI, 1.22–1.96; P = .0003). Having shown that the IL1RN rs419598 C/C genotype was associated with decreased mortality in men (Table 3), we next examined the age-dependence of the IL1RN effect. Figure 1B demonstrates that the rs419598 C/C SNV was associated with decreased age-related mortality in men, shown as a continuous variable (adjusted OR, 0.38; 95% CI, .15–.96; P = .04). A similar analysis in women showed no age-associated benefit of the rs419598 C/C SNV on mortality. Figure 1C and Table 4 examine mortality in men by decade. As shown, in patients under 74 years, the SNV C/C rs4198598 was associated with significantly decreased mortality (P = .001) for each decade. The combined decreased mortality for C/C patients ≤74 years of age was significant (4.1% vs 10.8%; OR = 0.37; 95% CI = .13–.84; P = .034; Table 5). However, this decreased mortality was observed only in male patients ≤74 years, whose mortality was reduced by 80% (3.1% vs 14%; OR = 0.20; 95% CI = .05–.71; P = .030; Table 5). Indeed, in the 64–75 age group, mortality in CT/TT patients was 22.2% (versus 7.1% in C/C carriers; P = .001; Figure 1C and Table 4). Decreased mortality of C/C in patients ≤74 years of age was associated with significantly lower inflammatory biomarker levels (IL-1β, IL-2, IL-6) than the C/T-T/T group (Table 5).

Table 4.

Mortality as the Percentage by Decade for IL1RN rs419598 C/C and C/T-T/T Variant Carriers Among Men

Age, y	C/C	C/T-T/T	All
19–44	0.0	3.3	3.0
45–54	0.0	7.7	7.4
55–64	3.8	14.5	13.6
65–74	7.1	22.2	21.7
75–80	20.0	21.7	21.5
80 and above	36.4	37.0	36.9

Table 5.

Age Dependence of the Association of the IL1RN rs419598 C/C Genotype With Decreased Mortality in Men

	≤74 y All IL1RN rs419598 (n = 1821)			≥ 75 y All IL1RN rs419598 (n = 665)
	C/C (n = 122)	C/T-T/T (n = 1799)		C/C (n = 53)	C/T-T/T (n = 612)
	Mean (SD)	Mean (SD)	FDR P Value	Mean (SD)	Mean (SD)	FDR P Value
Demography
Age, y	50.93 (15.12)	53.63 (14.95)	.0581	82.51 (5.85)	83.68 (5.63)	.2472
BMI	30.39 (6.46)	35.34 (145.30)	.1589	29.79 (7.68)	27.69 (7.64)	.1986
Sex, male, No. (%)	64 (52.5%)	1013 (56.3%)		26 (49.0%)	276 (45.1%)	.6807
Inflammation markers
Lab IL-1Ra	2281 (1448)	2676 (2030)	.4110	5378 (3622)	3591 (2969)	.2661
Lab IL-1β	5.91 (2.87)	6.79 (9.32)	.3140	5.21 (0.54)	7.05 (11.38)	.1609
Lab IL-2	5.36 (2.23)	7.47 (26.53)	.1902	5.09 (2.03)	6.93 (12.72)	.2123
Lab IL-6	20.72 (26.42)	45.50 (185.73)	.0074	31.03 (51.78)	38.57 (153.90)	.6061
IL-1β, max	5.41 (0.84)	7.67 (16.02)	.0170	5.54 (0.75)	7.29 (11.53)	.1750
IL-2, max	4.71 (1.98)	8.31 (48.19)	.1978	4.46 (2.40)	7.22 (15.36)	.1609
IL-6, max	59.20 (120.52)	151.00 (590.12)	.0067	67.16 (135.40)	76.24 (216.48)	.7472
Lab CRP	102.00 (66.61)	121.58 (92.37)	.0578	90.73 (60.16)	101.59 (81.97)	.3259
CRP, max	144.41 (98.66)	166.24 (113.23)	.1371	127.62 (82.73)	142.02 (102.33)	.3259
Lab procalcitonin	1.17 (6.12)	1.22 (6.94)	.9569	0.19 (0.36)	0.84 (3.44)	.0000
Lab ferritin	1553.08 (2408.01)	1542.80 (2545.53)	.9766	788.06 (1019.44)	1088.09 (2549.49)	.2123
Lab D-dimer	612.65 (967.64)	1266.79 (3846.52)	.0006	945.09 (1869.27)	1817.77 (4671.09)	.0800
D-dimer, max	1978.37 (4203.13)	3790.69 (7908.56)	.0065	2421.82 (3987.26)	3372.98 (6787.38)	.2514
Mortality, No. (%)	5 (4.1)	195 (10.8)	.37 (.13–.84); 0.034	15 (28.0)	182 (29.7)	.90 (.46–1.67); 0.739
Mortality, men only, No. (%)a	2 (3.1)	142 (14.0)	.21 (.03–.68); 0.030	7 (26.9)	88 (31.9)	.78 (.29–1.84); 0.582
Mortality, women only, No. (%)a	3 (5.2)	53 (6.7)	.79 (.19–2.24); 0.696	8 (29.6)	94 (28.0)	1.04 (.39–2.48); 0.937

Summary of demographics and inflammation biomarkers by genotype. The BMI is the weight in kilograms divided by the square of the height in meters. Biomarkers data are presented as mean (SD). Adjusted P values control for FDR at 5% using Benjamin-Hochberg criteria within each subgroup of biomarkers.

Abbreviations: BMI, body mass index; CRP, C-reactive protein; FDR, false discovery rate; IL, interleukin; Lab, routine laboratory tests on admission; Max, maximum values of inflammation markers during hospitalization.

^aFor mortality, comparison between genotype odds ratio (adjusted odds ratio [95% confidence interval] P value) derived from multivariable logistic regression adjusting for age, sex, and BMI; and for age and BMI for sex-specific mortality.

Figure 1B and 1C, Table 4, and Table 5 also illustrate that the survival benefit associated with the C/C rs419598 genotype was not observed for all patients over the age of 75 years. To further assess this finding, we compared biomarkers and mortality between ≤74 and ≥75 years of age according to IL1RN rs419598 genotypes (Table 5). Elevations of inflammatory cytokines did not differ in both age groups. However, as expected, there was a marked increase in age, sex, and BMI-adjusted mortality in patients ≥75 years of age versus ≤74 years (29.3% vs 10.4%; OR, 0.28; 95% CI, .23–.35; P <.0001; Table 5). The higher mortality was associated with significantly increased preexisting comorbid risk conditions among the older age group, including diabetes, coronary artery disease, heart failure, chronic lung disease, and cancer (Supplementary Table 6).

We examined IL1RN rs419598 C/C SNV genotype frequencies and comorbidities in patients above and below 75 years. The frequency of the C/C rs419598 genotype did not differ between those under versus those over 74 (6.4% vs 8.0%, P = .75). The C/C IL1RN SNV among those ≤74 years was associated with decreased IL-1β and IL-6, but not IL-2 levels. Mortality did not differ between the C/C and C/T-T/T cohorts (28% vs 29.7%) among patients 75 years and older (Table 5).

IL1RN SNVs and Outcomes in Black Versus White Patients

We next compared IL1RN SNVs and SARS-CoV-2 outcomes in our population's self-identified white and black patients. Supplementary Table 7 shows no difference in either plasma elevations of inflammatory biomarkers or the mortality between white and black patients. As has been reported, the frequency of the C/C rs419598 genotype was lower in black patients (1.1%) than in white patients (8%), comparable to values reported in the 1000 Genomes Database [15] (Ensembl GRCh37 release 103, February 2021, EMBL-EBI) (Supplementary Table 8). The frequency of 1 or 2 copies of the CTA haplotype was 10.2% in white patients and approximately 1% in black patients. Of interest, plasma IL-1Ra levels were lower in black patients than white patients (Supplementary Table 7), as reported [16].

DISCUSSION

These studies of 2589 hospitalized patients with SARS-CoV-2 demonstrate that the IL1RN CTA haplotype and its rs419598 C/C SNV were associated with the attenuation of the hyperinflammation that characterizes severe infection. Our data also show that in male patients ≤74 years of age, the rs419598 C/C SNV was associated with a significant decrease in mortality. These findings are concordant with our prior report on rheumatoid arthritis carriers of the IL1RN CTA haplotype, who exhibited lower disease activity scores, plasma CRP, and IL-6 in association with increased IL-1Ra [7]. Our observations in rheumatoid arthritis are among multiple prior reports that demonstrate genetic polymorphisms of the IL1RN gene modulate systemic inflammation in diverse inflammatory syndromes [17–20]. This is the first study that extends such observations to the hyperinflammation in SARS-CoV-2, which merits further investigation. Although the prevalence of the rs419598 C/C SNV is uncommon (7%–8%), when present its effects on the inflammatory response and mortality in men were profound, suggesting that activation of the IL-1 pathway plays a significant role in the pathogenesis of CRS in severe SARS-CoV-2 infection. In this regard, we note that important insights that result from the identification of genes of low frequency are not uncommon in the study of genetic influences in disease, where an identified rare mutation in a small population reveals an underlying pathogenic pathway for the larger affected population, as in the discovery of the CFTR gene mutation in the Amish population that led to a better understanding of the molecular basis of cystic fibrosis [21].

Our findings are supported by the 2013 study of Meyer et al, who found that the IL1RN variant rs315952, an SNV of the CTA haplotype here reported, is associated with a lower risk of ARDS and increased plasma IL-1Ra [14]. The authors used a 50,000 single-nucleotide polymorphism array in 3 separate ARDS populations to assess the risk of ARDS in critically ill patients. A total of 12 SNVs within 10 genes demonstrated an association with ARDS. Only the IL1RN SNV, rs315952C, met the predetermined level of statistical significance. We used a candidate IL1RN gene approach based on our prior studies of rheumatoid arthritis. We found that the same rs315952 SNV identified by Meyer et al is 1 of the 3 SNVs of the CTA haplotype reported here to attenuate the hyperinflammation of SARS-CoV-2 [14]. In our studies, the rs3155952 SNV, like CTA and rs419598, was associated with significantly decreased plasma IL-1β and maximum IL-1β (P < .001) (Supplementary Table 3).

IL-1β is considered to be among the most biologically active cytokines found in the lungs of patients with ARDS [22, 23]. Sefik et al reported that inflammasome activation in SARS-CoV-2–infected macrophages causing the release of IL-1, was a key driver of COVID-19 pathology [24]. Analysis of the cytokines contributing to CRS indicated that the anti-inflammatory cytokines IL-10 and IL-1Ra were significantly more elevated early in severe COVID-19 than the more frequently characterized cytokines such as IL-6, IL-1β, and TNF-α [25, 26]. Zhao et al speculated that the early elevation of IL-1Ra is a mechanism to control hyperinflammation responses during the early stages of immune activation [25].

An intriguing finding of this study is that IL1RN rs419598 SNV was associated with decreased mortality only in male patients. This is notwithstanding the finding that this genotype was associated with lower plasma inflammatory markers in both men and women (except for maximum IL-6, which did not differ between genotypes in women). Interestingly, the levels of IL-1Ra in women with the IL1RN rs419598 C/C genotype exceeded 4950 pg/mL, 2-fold higher than in women with the CT/TT genotype and 15-fold higher than normal.

How sex differences contribute to COVID-19 severity and mortality is still not clear. Interestingly, sex differences in IL-1Ra gene polymorphisms and IL-1Ra levels are well described [27–29]. Plasma IL-1Ra levels have been reported to be higher in men, while the in vitro production of IL-1Ra from isolated monocytes is significantly higher in women than in men [27, 28]. In addition to sex differences in regulating the IL-1 pathway, IL1RN polymorphisms act within an overall inflammatory SARS-CoV-2 milieu that may differ between men and women. For example, women have increased IFN-α secretion [30], and many immune and inflammation-association genes are X-linked, including but not limited to TLR7, TLR8, and IRAK1 [31]. In addition, the ACE2 gene resides in the X chromosome, and, thus, women theoretically have a double dose of ACE2, which may compensate for SARS-CoV-2–mediated loss of membrane ACE2 [31, 32]. Therefore, the interplay between IL1RN polymorphisms and other poorly understood sex differences may account for the differential effects of IL1RN genetic variants on mortality between men and women observed in our studies.

There are several limitations to our study. Because we employed low-pass whole-genome sequencing for genotype information, coverage may limit the ability to detect SNVs, limiting interpretations of haplotype and linkage disequilibrium. Because the data were retrieved from the EHR, cytokine data were unavailable for all patients. However, for the key proinflammatory cytokines of CRS (IL-1β, IL-6, IL-2), the number of individual patient values in the EHR was substantial, ranging from 642 to 1174 (Supplementary Table 1). Moreover, the IL1RN SNVs findings related to cytokines parallel the findings for conventional markers of inflammation (CRP, D-dimer, procalcitonin, ferritin), which were available for essentially all patients. These markers of SARS-CoV-2 hyperinflammation were significantly decreased in CTA and rs429598 carriers, reflecting reduced levels of the measured cytokine drivers such as IL-6 and IL-1β.

We note that the C/C rs419598 SNV is present in less than 10% of the white and Hispanic populations and <1% of those with African ancestry. Therefore, while this IL1RN SNV confers benefit in carriers and reveals insights into the role of the IL-1 pathway in severe SARS-CoV-2, there are likely other genetic determinants of disease severity SARS-CoV-2, including our own report regarding type 1 interferon genetic polymorphisms [33], as well as those noted above that may play a role regarding different outcomes in male versus female patients.

In summary, these studies demonstrate that the IL1RN CTA haplotype and rs419598 C/C genotype are associated with attenuation of the CRS in patients with severe SARS-CoV2 infection and reduced mortality in men. We show that concomitant with decreased proinflammatory cytokine production, the IL1RN CTA haplotype and rs419598 C/C SNV are associated with increased levels of its anti-inflammatory gene product IL-1Ra. Our data provide genetic evidence that activation of the inflammasome and the IL-1 pathway is proximal in the systemic cytokine inflammatory cascade. Its regulation by IL-1Ra, an endogenous anti-inflammatory protein, and potential crosstalk with IFN require further elucidation to advance the understanding and treatment of SARS-CoV-2 infection.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Author contributions. S. B. A., M. A., and A. C. contributed to the study concept and design. M. A., C. P., S. A., E. I., X. L., S. T., N. H., A. C., D.B., and S. B. A. contributed to the literature search, manuscript writing, and data interpretation. M. A., C. P., S. A., E. I., X. L., S. T., N. H., A. C., D.B., and S. B. A. participated in data collection and analysis. M. A., S. A., and X. L. made the figures and tables.

Financial support. This work was supported by the National Institutes of Health (NIH) (grant number R21-AR078466-01 to S. B. A.). The Immune Monitoring Laboratory, Division of Advanced Research Technologies, New York University Langone Health is partially supported by the NIH (grant number P30CA016087).

Author notes

M. A., C. P., and S. A. contributed equally.

Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.