Low serum transferrin predicts adverse outcome in COVID‐19 and is related to coagulation activity

To the Editor,

Transferrin is a hepatocyte‐derived protein which plays many different roles in the human organism. Its primary function is iron binding and transportation throughout the body, as free iron promotes oxygen radical formation with damage to tissues. Transferrin may furthermore have antimicrobial activity by sequestering free iron, as iron is essential for multiplication and pathogenicity of many pathogens. ¹ , ² As part of the nutritional immune response to invading pathogens, transferrin plasma concentration decreases in response to infection and inflammation. ² Alterations of iron homeostasis have been documented in the coronavirus disease 2019 (COVID‐19) ³ ; as these alterations have been associated with a poor prognosis, it was suggestive that transferrin may also serve as a good predictor of disease severity. ⁴ , ⁵ Such a notion was further supported by the finding that, in preclinical models, transferrin interferes with blood coagulation and that pathological coagulopathy was associated with adverse COVID‐19 outcomes. ¹ , ⁶

In this study, we analysed transferrin as a potential prognostic marker for severe COVID‐19 in hospitalized patients, as well as the relationship between transferrin levels and blood coagulation. We performed a single‐centre retrospective cohort analysis of patients hospitalized with COVID‐19 between February 2020 and March 2021. Blood samples were acquired within 3 days of hospitalization. Further information on the analysed laboratory parameters, clinical outcomes and statistical analysis are reported in the Supporting Information.

We analysed 621 hospitalized patients with polymerase chain reaction‐confirmed severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) infection: 247 women (39.8%) and 374 men (60.2%). Baseline characteristics are depicted in Table 1.

TABLE 1.

Baseline characteristics of all patients and according to the transferrin cut‐off.

	Total cohort	Transferrin <200 mg/dL	Transferrin ≥200 mg/dL	p‐Value a	n b
	n = 621	n = 410	n = 150
	Median (IQR) or n (%)	Median (IQR) or n (%)	Median (IQR) or n (%)
Demographic characteristics
Age, years	69 (55–78)	71 (59–79)	61 (47–75)	<0.001
BMI, kg/m2	26.3 (23.8–29.6)	26.0 (23.4–29.1)	26.4 (24.2–29.7)	0.106
Clinical characteristics
Temperature, °C	37.8 (36.9–38.5)	37.8 (37.1–38.6)	37.5 (36.6–38.3)	<0.001
SpO2, %	92 (88–94)	91 (88–94)	93 (90–96)	0.023
O2 requirement, L	2 (0–4)	2 (0–4)	1 (0–2)	<0.001
WHO score	4 (4–5)	4 (4–6)	4 (3–4)	<0.001
Hospitalization, days	9 (6–15)	10 (6–17)	8 (6–13)	0.010
ICU admission	152 (24.5%)	127 (31.0%)	20 (13.3%)	<0.001
Death during hospital stay	87 (14.0%)	68 (16.6%)	8 (5.3%)	<0.001
Comorbidities and risk factors
Cardiovascular disease	269 (43.3%)	187 (58.6%)	66 (47.5%)	0.028
Arterial hypertension	228 (36.7%)	160 (50.2%)	53 (38.1%)	0.018
Diabetes mellitus	123 (19.8%)	78 (24.3%)	37 (26.6%)	0.844
Chronic kidney disease	65 (10.5%)	45 (14.1%)	15 (10.8%)	0.162
Malignancies	60 (9.7%)	46 (14.4%)	12 (8.6%)	0.089
COPD	42 (6.8%)	28 (8.8%)	11 (7.9%)	0.761
Bronchial asthma	26 (4.2)	12 (3.8%)	12 (8.6%)	0.031
Nicotine abuse	16 (2.6%)	12 (3.8%)	3 (2.2%)	0.565
General laboratory parameters
Creatinine, mg/dL	0.97 (0.79–1.22)	0.99 (0.80–1.31)	0.90 (0.76–1.11)	0.003	616
AST, U/L	36 (26–54)	39 (27–58)	31 (24–44)	<0.001	613
ALT, U/L	25 (17–42)	25 (17–45)	24 (17–38)	0.352	566
ALP, U/L	67 (52–87)	66 (51–88)	69 (57–83)	0.414	561
HbA1c, %	6.1 (5.7–6.6)	6.1 (5.8–6.6)	6.0 (5.7–6.7)	0.160	522
Thrombocytes, 109/L	181 (148–241)	180 (145–239)	183 (156–242)	0.406	618
Haemoglobin, g/L	133 (120–145)	131 (117–143)	138 (127–150)	<0.001	618
Haematocrit, L/L	0.386 (0.349–0.418)	0.378 (0.344–0.409)	0.405 (0.373–0.432)	<0.001	618
aPTT, s	32 (29–36)	33 (29–36)	30 (27–34)	<0.001	579
INR	1.0 (0.9–1.1)	1.0 (0.9–1.1)	1.0 (0.9–1.0)	0.080	520
AT III, %	80 (73–89)	80 (72–87)	88 (79–96)	0.001	426
D‐dimer, μg/L	1111 (574–1971)	1219 (699–2318)	583 (411–1177)	<0.001	182
Fibrinogen, mg/dL	494 (381–583)	518 (425–612)	392 (336–515)	<0.001	331
Inflammatory biomarkers
CRP, mg/dL	5.07 (2.02–10.15)	6.47 (2.89–11.90)	2.24 (0.86–5.52)	<0.001	609
IL‐6, ng/L	35.7 (15.5–75.5)	43.9 (17.7–89.3)	22.5 (8.5–44.2)	<0.001	525
PCT, ng/mL	0.10 (0.06–0.27)	0.13 (0.07–0.35)	0.07 (0.05–0.12)	<0.001	562
Neopterin, nmol/L	46.8 (30.8–72.1)	54.5 (37.0–78.1)	36.4 (22.7–52.7)	<0.001	396
Ferritin, μg/L	460 (250–1.079)	649 (323–1.371)	261 (134–406)	<0.001	563
Leucocytes, 109/L	5.8 (4.5–7.8)	5.9 (4.5–8.1)	5.4 (4.4–7.1)	0.074	483
Lymphocytes abs, 109/L	17.8 (9.8–27.3)	15.7 (8.7–24.3)	23.8 (15.8–30.9)	<0.001	128

Note: p < 0.05 marked in bold.

Abbreviations: ALP, alkaline phosphatase; ALT, alanine aminotransferase; aPTT, activated partial thromboplastin time; AST, aspartate aminotransferase; AT III, antithrombin III; BMI, body mass index; COPD, chronic obstructive pulmonary disease; CRP, C‐reactive protein; GFR, estimated glomerular filtration rate; HbA1c, glycated haemoglobin; ICU, intensive care unit; IL‐6, interleukin 6; INR, international normalized ratio; IQR, interquartile range; O₂, oxygen; PCT, procalcitonin; SpO₂, peripheral capillary oxygen saturation; WHO, World Health Organization.

^a Kruskal–Wallis test or Pearson's chi‐squared test.

^b Number of patients with available laboratory marker.

Transferrin levels were available from 560 patients upon hospital admission (90.2%). We separated the patient cohort into subgroups with transferrin below and above 200 mg/dL, which is the lower threshold for normal transferrin levels in a healthy European population. ¹ Median length of stay declined with higher transferrin levels (rs = −0.159, p < 0.001). Necessity for intensive care unit (ICU) admission also decreased (Figure 1A) and mortality declined with higher transferrin levels (Table 1). Low transferrin levels predicted a higher risk for ICU admission (odds ratio [OR] 0.391 [95% confidence interval, CI 0.171–0.897], p = 0.027) and in‐hospital death (OR 0.342 [95% CI 0.124–0.936], p = 0.037) independent of sex, age, C‐reactive protein (CRP) and leucocyte count in logistic regression analysis. Patients with reduced transferrin levels (<200 mg/dL) had a 2.9‐fold higher risk for ICU admission (OR 2.917 [95% CI 1.743–4.883], p < 0.001) and 3.5‐fold higher risk of in‐hospital death (OR 3.529 [95% CI 1.653–7.533], p = 0.001, Figure 1B —Kaplan–Maier curve) compared to those with transferrin levels ≥200 mg/dL.

FIGURE 1.

Impact of transferrin and aPTT levels upon admission on risk for ICU admission or death. Patients who were transferred to ICU presented more often with transferrin levels <200 mg/dL (p > 0.001) (A) and an aPTT ≥38 s (C). Reduced transferrin levels (B) and a prolonged aPTT (D) were related to an increased mortality risk. aPTT, activated partial thromboplastin time; ICU, intensive care unit.

A prolonged activated partial thromboplastin time (aPTT) (≥38 s, based on our laboratory's upper reference limit) indicating that blood takes longer to clot was related to a 1.7‐fold higher risk for ICU admission (OR 1.698 [95% CI 1.025–2.815], p = 0.040, Figure 1C) and a 2.3‐fold higher risk for in‐hospital death (OR 2.284 [95% CI 1.317–3.960], p = 0.003, Figure 1D —Kaplan–Maier curve) independent of sex, age, CRP and leucocyte count in logistic regression analysis. Median length of stay increased with a higher aPTT (rs = 0.162, p < 0.001). The effect of further parameters on in‐hospital mortality and ICU admission is synthetized in the Supporting Information.

To investigate associations between different markers, we performed Spearman rank correlation analysis. Low transferrin was related to inflammation reflected by higher CRP (rs = −0.527, p < 0.001, Figure S1A), interleukin‐6 (IL‐6) (rs = −0.363, p < 0.001, Figure S1B) and leucocyte count (rs = −0.194, p < 0.001). Transferrin was further related to the coagulation markers aPTT (rs = 0.249, p < 0.001, Figure S2A), international normalized ratio (INR) (rs = −0.134, p = 0.002), antithrombin III (AT III) (rs = 0.337, p < 0.001, Figure S2B), fibrinogen (rs = −0.460, p < 0.001), D‐dimer (rs = −0.436, p < 0.001) and von Willebrand factor (vWF) Serpin Peptidase Inhibitor, Clade C (Antithrombin), Member 1 (rs = −0.358, p = 0.009), while no correlation was found with platelet count (rs = 0.012, p = 0.770).

Inflammation was associated with prolonged aPTT and INR, lower AT III, and higher fibrinogen and D‐dimer. Elevated CRP was related to a prolonged aPTT (rs = 0.233, p < 0.001) and INR (rs = 0.090, p = 0.034), lower ATIII (rs = −0.158, p = 0.027) and higher fibrinogen (rs = 0.740, p < 0.001) and D‐dimer (rs = 0.431, p < 0.001). Higher IL‐6 was linked to a prolonged aPTT (rs = 0.322, p < 0.001) and INR (rs = 0.136, p = 0.003), lower ATIII (rs = −0.269, p < 0.001) and higher fibrinogen (rs = 0.433, p < 0.001) and D‐dimer (rs = 0.347, p < 0.001).

Coagulation markers were widely correlated among each other. Prolonged aPTT was linked to higher fibrinogen (rs = 0.169, p = 0.002), low AT III (rs = −0.264, p < 0.001), and lower platelet count (rs = −0.163, p < 0.001), while no correlation was found with D‐dimer (rs = −0.013, p = 0.859) and vWF (rs = −0.126, p = 0.373). INR correlated with AT III (rs = −0.300, p < 0.001) and vWF (rs = 0.305, p = 0.030), but not with fibrinogen (rs = −0.034, p = 0.533), D‐dimer (rs = 0.129, p = 0.079) and platelet count (rs = −0.033, p = 0.436). Fibrinogen further correlated with D‐dimer (rs = 0.260, p = 0.008) and platelet count (rs = 0.144, p = 0.007), while no correlation was found with ATIII (rs = −0.188, p = 0.277) and vWF (rs = 0.319, p = 0.099). Finally, D‐dimer correlated with ATIII (rs = −0.176, p = 0.043).

For a schematic representation of the aforementioned correlation between transferrin, coagulation markers and inflammatory parameters, refer to the heat map in the Supporting Information.

Our study aimed to evaluate transferrin as a potential prognostic biomarker in SARS‐CoV‐2 infections and to study its potential link to hypercoagulation. Transferrin is the central iron transport protein which is pivotal for cellular iron acquisition. Its expression in the liver is induced by iron deficiency whereas concentrations decline with iron overload or inflammation. ⁷

The latter refers to the role of iron as an essential nutrient for many microbes which need iron for central metabolic pathways, multiplication and pathogenicity and have thus evolved multiple mechanisms to acquire or steal the metal from the infected host. ¹ , ² The host's immune system reacts to infection by reducing the amounts of available free iron in an attempt to limit pathogen growth. ⁸ Transferrin plays a role in the innate immune system not only by depriving certain microorganisms of iron but also by delivering iron to immune cells for their metabolic purposes and functionality. ² , ⁹ Therefore, lower levels of transferrin, as observed under infection or inflammatory conditions, limit iron access to circulatory pathogens but also negatively impact metabolic pathways and cellular metabolism in the host. ¹⁰ As a side effect, the ability of the host to bind and neutralize free iron is diminished upon reduction of circulatory transferrin.

As a consequence, anaemia has been associated with higher in‐hospital mortality in COVID‐19. ⁴ In our cohort, patients with the lowest transferrin levels had the highest mortality and necessity for ICU admission, potentially reinforcing the notion that abundance of uncontrolled free iron may lead to worsening of clinical outcomes.

In recent years, transferrin has further been investigated as a modifier in the coagulation process ⁶ because it may interfere with antithrombin/Serpin Family C, Member 1 (SERPINC1)‐mediated inhibition of coagulation proteases. ⁶ Previous studies on mouse models showed transferrin to be procoagulant by interacting with thrombin/Factor XIIa and blocking the inactivation effect of AT III on coagulation proteases. ⁶ A positive association of transferrin with hypercoagulation could not be confirmed in our study, in which, in contrast, low transferrin was associated with hyperfibrinogenaemia, elevated D‐dimer and vWF, which are factors related to prothrombotic states. Our clinical data are in line with another recent investigation, demonstrating that increased transferrin levels in people living at high altitudes are not associated with an increased risk of thrombosis. ¹¹ However, plasma transferrin is normally bound to fibrinogen with very little free transferrin in circulation being an additional explanation. ⁶

The correlation between transferrin and blood clotting in our study is not entirely clear, as lower levels of transferrin were conversely associated with lower AT III and prolonged aPTT and INR, typically observed in prolonged clotting time. ¹² Actually, transferrin elevation was related to a higher thrombosis risk. ⁶ Under hypoxic or iron‐deficient conditions, transferrin expression is elevated via enhanced hypoxia‐inducible factor signalling. ¹³ In patients with severe COVID‐19, one might expect hypoxia with consequently enforced transferrin expression. However, transferrin was significantly lower in our study suggesting that inflammation‐related transferrin suppression might overwhelm the hypoxia‐related upregulation. ¹⁴ , ¹⁵ Since prolonged coagulation was related to hyperfibrinogenaemia and lower AT III to more pronounced inflammation, this finding could thus be related to increased inflammation‐related consumption of coagulation factors. ¹²

One of the limitations of our study was its single‐centre design, which may have resulted in regional biases and limited patient diversity. Further individual patient diseases and fluctuations in the demographics of patients hospitalized during the study period may have introduced additional unmeasured confounders. Specifically, liver disease was not assessed in this cohort, and since transferrin is synthetized in the liver, this may represent an unmeasured confounder. During the observed period of time, various therapeutic options and vaccinations against COVID‐19 became available, potentially influencing disease severity, complications and clinical outcomes thus possibly introducing further unmeasured confounders in our data analysis.

To summarize, this retrospective cohort analysis found a significant association of low transferrin levels upon hospital admission with an increased length of stay, ICU admissions and in‐hospital mortality in patients infected with SARS‐CoV‐2. Reduced transferrin levels were associated with more advanced inflammation and coagulation activity which contrasts with preclinical data describing an association of high transferrin and hypercoagulation.

AUTHOR CONTRIBUTIONS

Conceptualization and methodology: GW. Software and formal analysis: LL and FRB. Investigation and data curation: LL, FRB and LT. Resources: RB‐W and GW. Writing—original draft preparation: LL and FRB. Writing—review and editing: LT, RB‐W and GW. Supervision: RB‐W and GW. All authors have read and agreed to the final version of the manuscript.

FUNDING INFORMATION

This research did not receive any specific grant from funding agencies in the public, commercial or not‐for‐profit sectors.

CONFLICT OF INTEREST STATEMENT

The authors declare that they do not have a conflict of interest.

ETHICS APPROVAL STATEMENT

The study conformed to the ethical principles outlined in the Declaration of Helsinki and was approved by the ethics committee of the Medical University of Innsbruck (decree number EK‐1167/2020).

PATIENT CONSENT STATEMENT

Because of the retrospective design, need for informed consent was waived by the local ethics committee.

Supporting information

Data S1.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.