The hallmark of severe Covid-19, a recently described systemic disease caused by severe acute respiratory coronavirus 2 (SARS-CoV-2), is the acute activation of the innate immune system associated with a prothrombotic state.¹ The clinical spectrum of Covid-19 ranges from asymptomatic cases to respiratory failure. Inflammatory cytokines, especially interleukin (IL) 6, IL1β, tumor necrosis factor alpha (TNFα), influence iron metabolism mainly triggering the production of hepcidin, a liver-secreted antimicrobial peptide-hormone that limits iron availability by blocking the iron exporter ferroportin.² High hepcidin during systemic inflammation, by reducing serum iron, impairs hemoglobin synthesis and red blood cell production leading to anemia of inflammation.² In addition, iron deficiency affects the functionality of the immune system and the sensing of hypoxia.³ Few data are available on the possible contribution of altered iron homeostasis on Covid-19.^3-5 In a retrospective study involving 50 Covid-19 patients, serum iron was found extremely low in most cases and a predictor of mortality.⁴ In addition, alterations of iron metabolism was associated with hypoxemia in severe Covid-19 patients in the intensive care unit (ICU).^{5, 6}

Hepcidin may represent the link between iron deficiency and Covid-19 severity. However, the relative contribution of hepcidin upregulation to the disrupted iron homeostasis in Covid-19 remains to be determined.

In the present study, we analyzed hepcidin levels in 111 Italian Covid-19 patients admitted to San Raffaele University Hospital in Milan. The cohort had a median (IQR) age of 57.6 (48.5-66.3) years, and included prevalently males (64%). A full demographic and clinical characterization of the cohort is reported elsewhere (Rovere-Querini et al., manuscript in preparation). Blood samples were obtained at admission, plasma immediately retrieved and frozen until analyzed for the concentration of iron, hepcidin, the major regulator of iron homeostasis (ELISA kit from Intrinsic LifeSciences), proinflammatory markers as C-reactive protein (CRP) and ferritin, and cytokines with a role in hepcidin modulation, as interleukin IL6, IL1β, TNFα and interferon gamma (IFNγ). The relative results are in Table S1. Iron concentration was below normal range in 93.7% of patients (Figure S1A). Conversely hepcidin levels were significantly increased in 61.3% of patients (Figure S1B). However, considering that hypoferremia suppresses hepcidin expression,² even normal hepcidin is inappropriately high in most cases.

As shown in the correlation with biochemical parameters (Figure S1C), patients with higher hepcidin levels were significantly older and had higher concentrations of markers of inflammation (CRP and ferritin) and cell damage (AST and LDH). Interestingly, plasma hepcidin levels were correlated with those HT of creatinine, possibly because hepcidin undergoes renal excretion. Negative H correlations were observed with the severity of respiratory failure, as reflected by the PaO₂/FiO₂ ratio. Iron levels, already low at admission, were comparable between males and females (Figure S1D), whereas hepcidin levels significantly correlated with gender, with males having higher values than females in our cohort of Covid-19 patients (Figure S1E). Hepcidin synthesis is mainly driven by inflammation and iron: in our cohort, it significantly correlated with markers of inflammation but not with plasma iron levels. Overall the large majority of hospitalized Covid-19 patients was hypoferremic and had inflammation-dependent hepcidin upregulation of variable entity.

We then stratified patients according to PaO₂/FiO₂ ratio >300 mm Hg (normoxemic) vs PaO₂/FiO₂ ratio ≤300 mm Hg (hypoxemic). Patient clinical features of the two groups are in Table S2. Patients in the hypoxemic group were prevalently males (76.2%) and significantly older (Figure S2A). Iron levels, although lower in hypoxemic females than in hypoxemic males, were similar in tHOhe same gender of the normoxemic and hypoxemic groups (Figure S2B). Hepcidin (Figure S2C), CRP (Figure S2D), LDH and AST (not shown) were all significantly higher in the hypoxemic group. Surprisingly, proinflammatory cytokines such as IL6 (Figure S2E), IL1β (Figure S2F), TNFα (Figure S2G) and IFNγ (Figure S2H) levels were comparable between groups.

Iron metabolism is strongly influenced by gender, with iron and hepcidin levels normally lower in females than in males. Due to the different distribution of males and females in the two groups, and the potential effects of differences in iron and hepcidin levels we repeated the analysis according to gender. While hepcidin levels were lower in females than in males in the normoxemic group, they were significantly higher, reaching male values, in the hypoxemic patients (Figure S2C). This indicates a more reactive inflammatory response in terms of hepcidin and hypoferremia in female patients with Covid-19, as previously reported for other infectious diseases. Thus the hepcidin increase in the total cohort was influenced by the strong increase of hepcidin in females. In the hyOpoxemic group, CRP was increased in both genders (Figure S2D), whereas proinflammatory cytokines levels remained unchanged (Figure S2E-H).

Seven weeks after admission, 89 (80%) patients with a mean age of 55.80 years survived, while 22 (20%) patients, mainly males with a mean age of 65.39 years, were dead (Table S1). The concentration of hepcidin at admission was significantly higher in patients who did not survive (Figure S3A), whereas iron levels showed only a trend towards a decrease in the group of patients deceased at 7 weeks (Figure S3B). Interestingly, the only patients with iron levels within the normal range were in the survivor group.

Kaplan-Meier curves depict the difference in survival between patients with hepcidin levels above the median (361.9 ng/mL) and those with hepcidin normal or below the median (Figure 1A). The same analysis (Figure 1B) was also performed to explore differences in survival between patients with decreased iron (below the median, 27.7 μg/dL) and patients with normal or increased iron (above the median).

Interestingly, hepcidin measured at admission predicts the clinical outcome (Figure 1A and S3A) as other validated parameters, such as age (Figure S3C) and CRP (Figure S3D). On the contrary, iron (Figure 1B and S3B), IL6 (Figure S3E), IL1β (Figure S3F) and IFNγ (Figure S3H) were similar in the two groups, and TNFα showed only a trend towards an increase in the non-survivor group (Figure S3G). The contribution of females, who have low iron and hepcidin levels, is unlikely to represent a confounder, since the proportion of females was similar in non-survivors (32%) and survivors (36.8%). With regression tree (RT) analysis, hepcidin, age and CRP emerged as robust classifiers of patients into risk groups for death (Figure S4). Specifically, patients with hepcidin levels below 394 ng/mL had the lowest probability of dying. Patients with high hepcidin levels differed in the probability of poor outcome depending on age and CRP levels at admission. The area under the ROC curve (ROC-AUC) (95% CI) for the RT was 0.79 (0.68-0.91). Overall these data demonstrate that hepcidin is a clinical biomarker of Covid-19 severity.

Finally we analyzed iron and hepcidin levels in the subset of the most severely affected patients admitted to the ICU (Table S1). Eighteen out of 36 (50%) ICU patients died. Since the distribution of females in the two subgroups was unbalanced (one female in the survivors and five in the non survivors), we analyzed only males. Age (Figure 1C), PaO₂/FiO₂ ratio (not shown), CRP (Figure 1D), and proinflammatory cytokines (Figure S5) were comparable between ICU patients who survived or not. Also plasma iron levels did not change between the two groups (Figure 1E), likely because of the low basal levels. On the contrary, hepcidin concentration was significantly increased in males who did not survive (Figure 1F), confirming the predictive role of hepcidin for death in severely affected Covid-19 patients.

Assessment of various biomarkers for the ability to predict the outcome of our cohort of Covid-19 points out an important role for hepcidin and iron analyzed at the time of hospital admission. Advanced age, low PaO₂/FiO₂ ratio, increased CRP and hepcidin and hypoferremia all characterize non-survivors. The role of hepcidin is strengthened by the Kaplan-Meier survival curve and confirmed by the RT analysis, which identified hepcidin as the most important predictor of death among the others recognized predictors. On the other hand, iron levels do not affect survival, likely because of the uniformly low levels in all patients.

Interestingly, limiting the analysis to critical patients in ICU, high hepcidin predicts mortality, independently of age, lung function, inflammation and tissue damage. Overall our data suggest that in Covid-19 hepcidin can be considered a marker of morbidity and outcome, of special value for severely compromised patients in ICU. Further studies are necessary to verify whether targeting the hepcidin axis may influence the disease outcome.

CONFLICT OF INTEREST

The authors declare no conflict of interest.