Serum IFN-γ and RNAemia temporal profiles as biomarkers of severe COVID-19 in solid organ transplant and immunocompetent patients

Dear editor,

Currently, the availability of SARS-CoV-2 vaccines, despite the great worldwide differences in vaccination rates,¹ have focused the impact of the pandemic in immunodepressed patients. Therefore, we read with interest the recent meta-analysis about COVID-19 vaccine in patients with solid malignancies² which found that 42% and 86% of patients achieved serological response after one and two doses, respectively. Other authors have found that the pooled odds ratio for developing anti-SARS-CoV-2 spike protein IgG was higher in the control group than in solid organ transplant (SOT) recipients.³ The infection by SARS-CoV-2 elicits an innate and specific cellular and humoral immune response. Interferons (IFN) are key in the innate immune response during the acute phase of the viral infection, as seen with plasmacytoid dendritic cells expressing high concentrations of types I and III IFNs in COVID-19 patients.^{4 , 5} In acute COVID-19 and convalescent patients, intracellular cytokine staining after stimulation with SARS-CoV-2 peptide pools, has showed significant IFN-γ increases in CD8 + T-cells, which is associated with viral elimination, and without differences in both phases of the disease.^{6 , 7}

In this context, to gain insight in the innate immune response in SOT recipients with COVID-19, compared with no SOT patients, we have assessed the IFN-α and IFN-ℽ serum levels and RNAemia at hospital admission and by days from the symptom's onset (DfSO; ≤1 to ≥15 days), as well as their association with unfavorable clinical outcomes (death and/or invasive mechanical ventilation [IMV]). With this aim, we conducted a multicentre prospective observational cohort study, including consecutive adult inpatients with confirmed COVID-19 (RT-PCR in nasopharyngeal swabs) and available samples for IFN-α/IFN-γ serum levels and RNAemia determinations (Methods, Supplementary material), from January 6th 2020 to August 13th 2021, followed until hospital discharge, death, or 30 days, whichever occurred first. The study was approved by The Ethics Committee of University Hospitals Virgen Macarena and Virgen del Rocío (C.I. 0771-N-20 and 0842-N-20).

Data were separately analyzed for SOT recipients and no SOT patients. In addition, we performed a matched cohort analysis in which patients undergoing SOT were paired with those from the no SOT cohort (1:2) according to their propensity score (PS), using callipers of a 0.01 standard deviation, to control for residual confounders. Mortality in the matched pairs was compared using Cox regression. IFN-α and IFN-γ levels were analyzed as discrete (undetectable and detectable) and continuous (pg/mL) variables. Multivariate Cox regression and logistic regression analysis were performed to identify factors independently associated with 30-day all-cause mortality and unfavorable clinical outcomes (Methods, Supplementary material).

Forty-seven (10.3%) SOT recipients and 408 (89.7%) no SOT patients (Supplementary Table 2) were recruited. The mean DfSO to hospital admission was 7.1 ± 4.3, without differences between groups. Undetectable IFN-α occurred in 8.5% and 13.5% (p = 0.36) of both groups, respectively (Fig. 1 A), independently of the DfSO. In SOT recipients, IFN-α levels were higher with <7 DfSO than with ≥7 DfSO (p = 0.015), with a decrease from 19.5 pg/mL to 1.4 pg/mL. In no SOT, IFN-α levels were higher than in SOT recipients with ≥7 DfSO (Supplementary Table 3). Undetectable IFN-γ was more frequent in SOT recipients than in no SOT patients (42.6% and 19.4%, p < 0.001) and this difference was higher with ≥7 DfSO (Fig. 1A). IFN-γ levels were similar over the different time-periods, both in SOT and no SOT, and without differences between groups (Supplementary Table 3), which is consistent with other studies.^{6 , 7} RNAemia was more frequent in SOT recipients (57.4%) than in no SOT patients (18.9%, p < 0.001) (Fig. 1A, Supplementary Table 2). In SOT recipients, RNAemia detection was independent of the DfSO, and in no SOT patients decreased with ≥11 DfSO (p = 0.014) (Supplementary Table 3). Mortality was higher in SOT recipients than in no SOT patients with ≥4 DfSO (Fig. 1A). In SOT recipients, mortality was not associated to the DfSO at admission; however, in no SOT patients, mortality was much higher in patients with ≤3 DfSO, decreasing to 0% in patients with ≥11 DfSO (p < 0.001) (Fig. 1A).

Fig. 1.

SOT recipients (n = 47) comparison with no SOT patients (n=408) regarding (A) undetectable IFN-α and IFN-γ serum levels, RNAemia, and mortality, by days from symptoms onset at hospital admission, and (B) Survival Kaplan Meier analysis of patients with and without undetectable IFN-γ serum levels and RNAemia detection.

In the PS matched cohorts (Table 1 ), SOT recipients showed higher prevalence of undetectable IFN-γ than no SOT patients (39.4% vs. 10.6%, respectively; p = 0.001), lower plasma IFN-α and IFN-γ levels in those with RNAemia (p = 0.013 and p = 0.001, respectively; Supplementary Fig. 1B), higher RNAemia detection (57.6% vs. 13.6%; p < 0.001) and mortality (27.3% vs. 4.5%; p = 0.003).

Table 1.

Comparison of solid organ transplant (SOT) recipients matched (1:2) with no SOT patients according to propensity score.^a

Variable	SOT recipients (n = 33)	No SOT patients (n = 66)	P value
Male sex	20 (60.6)	37 (56.1)	0.666
Age >70 years	4 (12.1)	21 (31.8)	0.033
Dyspnoea	15 (45.5)	36 (54.5)	0.394
SpO2 <95%	17 (51.5)	26 (39.4)	0.251
Neutrophil count >7500/μL	6 (18.2)	6 (9.1)	0.327
Lymphocyte count <1000/µL	18 (54.5)	30 (45.5)	0.394
C-reactive protein >100 mg/L	9 (27.3)	20 (30.3)	0.755
Ferritin >1000 ng/mL	6 (18.2)	10 (15.2)	0.699
D-dimer >600 ng/mL	29 (87.9)	48 (72.7)	0.087
LDH >300 IU/L	14 (42.4)	33 (50.0)	0.477
IFN-α undetectable	3 (9.1)	8 (12.1)	0.747
IFN-α (pg/mL)b	1.43 (0.60–22.01)	11.98 (3.24–23.11)	0.163
IFN-γ undetectable	13 (39.4)	7 (10.6)	0.001
IFN-γ (pg/mL)b	26.14 (0.00–240.96)	145.35 (40.00–330.96)	0.347
RNAemia positive	19 (57.6)	9 (13.6)	<0.001
RNAemia (log10 copies/mL)b	2.38 (2.12–3.19)	2.36 (1.92–2.98)	0.921
CCI ≥3	25 (75.8)	37 (56.1)	0.056
CURB-65 ≥2	12 (36.4)	12 (18.8)	0.057
WHO basal score 6–9c	3 (9.1)	4 (6.1)	0.683
IMV	10 (30.3)	5 (7.6)	0.003
Mortality at day 30	9 (27.3)	3 (4.5)	0.003
WHO final score 7-10c	13 (39.4)	7 (10.6)	0.001

Data are presented as No. (%). P values are calculated by Cox regression.

Abbreviations (in order of appearance): SpO₂, peripheral capillary oxygen saturation; LDH, lactate dehydrogenase; IFN, interferon; CCI, Charlson Comorbidity Index (30); CURB-65 (31), Severity Score for Community-Acquired Pneumonia; IMV, invasive mechanical ventilation.

^aVariables included in the propensity score were sex, dyspnea, SpO₂, neutrophil and lymphocyte counts, C-reactive protein, ferritin, D-dimer, and LDH.

^bMedian (IQR). P values are calculated by the Mann-Whitney U test.

^cSeverity rating according to the WHO Clinical Progression Scale (doi: 10.1016/S1473-3099(20)30483-7), ranged from 0 (not infected) to 10 (dead), of which scores 6–9 represent severe disease.

In SOT recipients, the multivariate logistic regression model selected RNAemia as predictor of unfavorable clinical outcome (Supplementary Table 6). Regarding no SOT patients, in the Cox regression multivariate analysis, 30-day all-cause mortality was associated with RNAemia and undetectable IFN-γ levels (Supplementary Table 7). In the Kaplan-Meier analysis, patients with RNAemia had lower survival, both in SOT (p < 0.0133) and no SOT (p = 0.001) groups (Fig. 1B). RNAemia has been associated with COVID-19 mortality.⁸ The present data confirm it, with a higher sample size and including SOT recipients, in which the RNAemia impact had not yet been analyzed. The Kaplan-Meier analysis also showed an association of undetectable IFN-γ with lower survival in SOT recipients (p = 0.048) (Fig. 1B). Our results, showing an association of undetectable IFN-γ in serum with mortality, support the protective role of the specific T-cells response.

The Kaplan-Meier analysis did not show association of undetectable IFN-α levels with the survival at 30 days, both in SOT and no SOT groups. It has been reported that inborn errors of type I IFN immunity accounts for life-threatening COVID-19 pneumonia⁹ and that autoantibodies against type I IFNs increased the infection fatality rate.¹⁰ However, we have not found association of serum undetectable IFN-α with unfavorable outcome, including immunosuppressed patients as the SOT recipients. A limitation is that we did not analyse IFN-stimulated genes to define a type I IFN signature nor interferon autoantibodies, because of our purpose was to identify easy-to-measure variables in the clinical setting.

In summary, the present results support the RNAemia and IFN-γ serum levels determinations, at hospital admission, in all adult COVID-19 patients, to guide their management and to assess the antiviral therapy efficacy in the case of RNAemia.

Declaration of Competing Interest

The authors have declared that no competing interests exist.