ABSTRACT
Liver disease is a leading cause of death worldwide. Nucleated red blood cells (NRBCs) are associated with high mortality and poor outcomes in patients with severe illnesses. However, the relationship between NRBCs and severe liver disease remains unclear. Potential confounding effects were managed using propensity score matching. The association between NRBCs and clinical outcomes in patients with liver disease was clarified using Cox proportional hazards regression analysis and smoothing splines. Differences in NRBCs between 30‐day survivors and non‐survivors within the pre‐matched cohort during the first 30 days after ICU admission were assessed using generalised additive mixed models. Compared to the 30‐day survivors, the 30‐day non‐survivors had significantly higher NRBC counts. Higher NRBC counts were significantly correlated with an augmented risk of 30‐day, 90‐day and in‐hospital mortality, with concurrently decreased hospitalisation durations. Inpatients with liver disease, progressive increases in the 30‐day mortality risk were associated with increased NRBC counts. The association between NRBCs and enhanced 30‐day mortality rates was consistent across stages and etiologies. Moreover, 30‐day non‐survivors experienced average daily increases in NRBC counts of 0.31% compared with 30‐day survivors. Elevated NRBC counts correlated with increased 30‐, 90‐day and in‐hospital mortality in patients with liver disease.
Keywords: all‐cause mortality, liver diseases, MIMIC‐IV database, nucleated red blood cells
Abbreviations
- AARC
APASL ACLF research consortium
- Alb
albumin
- ALT
alanine transaminase
- AST
aspartate transaminase
- BIDMC
Beth Israel Deaconess Medical Center
- BUN
blood urea nitrogen
- CECs
CD71+ erythroid cells
- CI
confidence interval
- COPD
chronic obstructive pulmonary disease
- Cr
creatinine
- GAMMs
Generalised Additive Mixed Models
- HIV
human immunodeficiency virus
- HR
hazard ratio
- ICU
intensive care unit
- INR
international normalised ratio
- Lac
lactate
- LOS hospital
length of hospital stay
- LOS ICU
length of ICU stay
- LRT
logarithmic likelihood ratio test
- Meld
model for end‐stage liver disease
- MIMIC‐IV
Medical Information Mart for Intensive Care‐IV
- MIT
Massachusetts Institute of Technology
- NRBCs
nucleated red blood cells
- PLT
platelet count
- PSM
propensity score matching
- SMD
standardised mean difference
- SOFA
Sequential Organ Failure Assessment
- SQL
Structured Query Language
- TBIL
total bilirubin
- WBC
white blood cell count
1. Introduction
The liver, located in the upper right part of the abdomen, is vital for bilirubin and nutrient metabolism, blood volume regulation, bile secretion, detoxification, mineral and vitamin storage, synthesis of prothrombin, fibrinogen and clotting factors and immunoregulation [1]. Despite its robust functionality, various factors, such as drugs, environmental toxicants, viruses and dietary components, can impair its function [2]. When liver damage occurs, lipid metabolism in hepatocytes is dysregulated, liver function is impaired and the immune system is activated, leading to various liver diseases [3, 4, 5]. Liver diseases include viral hepatitis, nonalcoholic fatty liver disease, alcoholic liver disease, cirrhosis, liver failure, autoimmune liver disease and hepatocellular carcinoma (HCC) [6]. Each of these conditions poses a unique challenge to liver health, potentially leading to progressive dysfunction and posing significant risks to overall well‐being [7]. Worldwide, liver diseases have become a prominent factor in illness and death rates [8], leading to an annual loss of two million lives, representing 4% of global deaths [9]. Therefore, elucidating the mechanisms underlying liver disease development and identifying potential early biomarkers is essential.
Nucleated red blood cells (NRBCs) are progenitor cells involved in mammalian erythropoiesis [10]. In healthy adult humans, NRBCs are not found in peripheral blood, as nuclear extrusion occurs prior to their release into the circulation [11]. However, under certain pathological conditions, particularly those associated with severe anaemia or stress, the bone marrow may accelerate erythropoiesis and prematurely discharge NRBCs into peripheral blood [12]. Hypoxemia of the arterial system may contribute significantly to the formation of NRBCs [13]. The process by which NRBCs enter the bloodstream from the bone marrow is unclear, but their appearance in adult peripheral blood is generally considered an abnormal finding and may indicate underlying haematological disorders, malignancies or acute hemolytic processes, which are also signs of disease severity and an increase in mortality [14, 15, 16]. For example, the presence of NRBCs has been proposed as an indicator of overall in‐hospital mortality in individuals with sepsis, septic shock and cardiovascular diseases [17, 18, 19]. In addition, Li et al. observed CD71+ erythroid cells (including NRBCs and reticulocytes) among extramedullary haematopoietic cells in the liver at the time of biliary atresia diagnosis. They also discovered that preemptive depletion of hepatic CD71+ erythroid cells with antibodies safeguarded neonatal mice from rhesus rotavirus‐induced biliary atresia [20]. Jie et al. reported a significant increase in the number of CD45+CD71+ erythroid cells in HCC tissues. These cells exert immunosuppressive effects on the tumour microenvironment and can act as important biomarkers for predicting HCC recurrence [21].
However, a systematic and comprehensive assessment of the potential of NRBCs as predictors of liver disease remains lacking. Liver diseases might be of particular importance in this regard because the liver is an extramedullary haematopoietic organ. Hence, in this study, we analysed data on patients with liver disease from the Medical Information Market for Intensive Care IV (MIMIC‐IV) (v2.0) database, focusing on the association between NRBCs and prognosis in patients with liver disease. This study may provide insights into the predictive value of NRBCs for clinical outcomes in patients with liver disease, as well as new approaches toward improving prognoses in this population.
2. Materials and Methods
2.1. Study Design
This retrospective cohort study used data from the MIMIC‐IV database, which contained information on various clinical parameters. Permission to access the MIMIC‐IV database was obtained (Record ID: 40060500). To safeguard patient confidentiality, all personally identifiable information was de‐identified, and the requirement for informed consent was waived. The study was approved by the institutional review boards of the Massachusetts Institute of Technology (MIT) and the Beth Israel Deaconess Medical Center (BIDMC).
2.2. Patient Selection
The inclusion criteria were as follows: (1) liver disease diagnoses, including liver failure, cirrhosis, hepatitis (alcoholic, autoimmune and viral), hepatorenal syndrome, liver necrosis, liver tumour, fatty liver, hepatic coma, liver abscess and liver fibrosis; and (2) available NRBC data. The exclusion criteria were as follows: (1) haematological malignancies and other haematological diseases, (2) solid carcinomas, (3) immunosuppression and (4) treatment with chemotherapy and corticosteroids.
2.3. Data Collection
Structured Query Language (SQL) and PostgreSQL were used to extract data. The demographic data included age, sex, ethnicity and body mass index (BMI). The disease severity was assessed using the model for end‐stage liver disease (MELD) score [22], Sequential Organ Failure Assessment (SOFA) score [23] and APASL ACLF research consortium (AARC) score [24] for each patient. Baseline laboratory values included the platelet count (PLT); levels of alanine transaminase (ALT), aspartate transaminase (AST), albumin (Alb), total bilirubin (TBIL) and blood urea nitrogen (BUN); international normalised ratio (INR), lactate (Lac) levels; white blood cell count (WBC); and creatinine (Cr) level, based on the highest values recorded. Comorbid conditions included hypertension, diabetes, congestive heart failure, cerebrovascular disease, chronic obstructive pulmonary disease (COPD), renal disease and sepsis. Baseline NRBC counts were measured within the first 24 h of ICU admission, with the highest value selected in cases of multiple measurements. Subsequent NRBC counts were performed at irregular intervals.
The outcome measures included 30‐day mortality, 90‐day mortality, in‐hospital mortality, length of hospital stay (LOS‐hospital) and length of ICU stay (LOS‐ICU).
2.4. Statistical Analysis
Continuous variables are expressed as mean ± standard deviation or median with interquartile range and were analysed using the Student's t‐test for a normal distribution or Kruskal–Wallis H test for a non‐normal distribution. All patients were categorised based on 30‐day mortality outcomes. To determine the links between baseline variables and 30‐day mortality, univariate Cox analysis was performed.
Propensity score matching (PSM) was utilised to account for variations between the low NRBCs group (< 1.0%) and high NRBCs group (≥ 1.0%) [25]. Propensity scores were derived from demographic information, scoring systems, laboratory data and comorbidities. One‐to‐one nearest neighbour matching was applied, with a calliper width of 0.02. The patient outcomes in the two groups were analysed in both the pre‐ and post‐matched cohorts.
To determine the independent effects of the baseline NRBC count on 30‐day mortality, we applied multivariate Cox regression analysis and smooth curve fitting using both unadjusted and adjusted models. Moreover, a segmented regression model and logarithmic likelihood ratio test (LRT) were used to examine the threshold effect between NRBC counts and 30‐day mortality [26]. The risk of 30‐day death in patients with liver disease was stratified by disease process (e.g., hepatitis, cirrhosis or liver failure) and aetiology (e.g., viral infection or alcohol), and the interaction of NRBCs with these stratified groups was assessed.
Generalised Additive Mixed Models (GAMMs) were used to visualise and quantify the differences in NRBC count changes between 30‐day survivors and non‐survivors within the pre‐matched cohort during the first 30 days after ICU admission. GAMMs were well suited for analysing longitudinal data and effectively handled unbalanced and unequally spaced observations.
Statistical analyses were performed using R and EmpowerStats software. Statistical significance was set at p < 0.05.
3. Results
3.1. Participants Selection
Overall, 22,430 adult patients with liver disease admitted to the ICU and with data in the MIMIC‐IV database (v2.0) were initially included. Of these, information on NRBCs was available for 1926 patients. We excluded patients with haematological malignancies and other haematological diseases (n = 93), solid carcinomas (n = 597) and immunosuppression (n = 163) and patients treated with chemotherapy drugs and corticosteroids (n = 52). Finally, 1021 patients were eligible: 603 participants in the 30‐day survivor group and 418 participants in the 30‐day non‐survivor group (Figure 1).
FIGURE 1.
Flow chart of the study.
3.2. Baseline Patient Characteristics
Table 1 shows the baseline characteristics of the eligible patients with liver disease. Among them, 418 (40.94%) died within 30 days. Compared with the 30‐day survival group, the 30‐day non‐survivor group had higher NRBC counts [1.00 (0.30–2.95) vs. 0.90 (0.20–1.30)], ages [61.89 ± 15.16 vs. 56.40 ± 15.34], MELD scores [28.64 ± 9.12 vs. 23.24 ± 9.79], SOFA scores [12.00 (9.00–15.00) vs. 0.90 (0.20–1.30)], AARC scores [10.69 ± 1.58 vs. 9.86 ± 1.65], AST levels [200.00 (78.00–1073.75) vs. 169.00 (59.00–557.50)], TBIL levels [2.80 (1.40–9.88) vs. 2.60 (1.00–5.75)], BUN levels [42.00 (27.00–67.75) vs. 30.00 (18.00–54.00)], INR[2.30 (1.70–3.40) vs. 1.80 (1.40–2.50)], Lac levels [7.19 ± 4.75 vs. 5.32 ± 2.97], WBC counts [16.35 (10.60–23.78) vs. 13.90 (9.30–20.20)], Cr levels [2.30 (1.42–3.60) vs. 1.60 (1.00–3.10)], and sepsis rates [367 (87.80%) vs. 459 (76.12%)] and lower PLT counts [90.00 (55.00–152.00) vs. 114.00 (66.00–179.00)] and Alb levels [2.70 ± 0.67 vs. 2.84 ± 0.56].
TABLE 1.
Baseline characteristics of total cohort, 30‐day survivors and 30‐day non‐survivors.
| Variables | Total (n = 1021) | 30‐day survivor (n = 603) | 30‐day non survivor (n = 418) | p |
|---|---|---|---|---|
| Demographic data | ||||
| Age (year) | 58.65 ± 15.49 | 56.40 ± 15.34 | 61.89 ± 15.16 | < 0.001* |
| Gender | ||||
| Male | 579 (56.71%) | 346 (57.38%) | 233 (55.74%) | 0.603 |
| Female | 442 (43.29%) | 257 (42.62%) | 185 (44.26%) | |
| Ethnicity | ||||
| White | 607 (59.45%) | 376 (62.35%) | 231 (55.26%) | < 0.001* |
| Black | 121 (11.85%) | 82 (13.60%) | 39 (9.33%) | |
| Other | 293 (28.70%) | 145 (24.05%) | 148 (35.41%) | |
| BMI (kg/m2) | 30.49 ± 6.60 | 30.50 ± 6.00 | 30.49 ± 7.38 | 0.994 |
| Score system | ||||
| Meld score | 25.45 ± 9.88 | 23.24 ± 9.79 | 28.64 ± 9.12 | < 0.001* |
| SOFA score | 10.00 (6.00–13.00) | 8.00 (6.00–11.00) | 12.00 (9.00–15.00) | < 0.001* |
| AARC score | 10.20 ± 1.67 | 9.86 ± 1.65 | 10.69 ± 1.58 | < 0.001* |
| Laboratory indicators | ||||
| PLT (K/μL) | 102.00 (59.00–169.00) | 114.00 (66.00–179.00) | 90.00 (55.00–152.00) | < 0.001* |
| ALT (IU/L) | 80.00 (30.00–403.00) | 80.00 (30.00–314.00) | 83.50 (32.00–502.75) | 0.802 |
| AST (IU/L) | 169.00 (68.00–759.00) | 169.00 (59.00–557.50) | 200.00 (78.00–1073.75) | < 0.001* |
| Alb (g/dL) | 2.78 ± 0.61 | 2.84 ± 0.56 | 2.70 ± 0.67 | < 0.001* |
| TBIL (mg/dL) | 2.60 (1.10–7.30) | 2.60 (1.00–5.75) | 2.80 (1.40–9.88) | < 0.001* |
| BUN (mg/dL) | 35.50 (20.00–59.00) | 30.00 (18.00–54.00) | 42.00 (27.00–67.75) | < 0.001* |
| INR | 1.90 (1.50–2.90) | 1.80 (1.40–2.50) | 2.30 (1.70–3.40) | < 0.001* |
| Lac (mmol/L) | 6.09 ± 3.91 | 5.32 ± 2.97 | 7.19 ± 4.75 | < 0.001* |
| WBC (K/μL) | 14.70 (9.90–21.80) | 13.90 (9.30–20.20) | 16.35 (10.60–23.78) | < 0.001* |
| Cr (mg/dL) | 1.90 (1.10–3.40) | 1.60 (1.00–3.10) | 2.30 (1.42–3.60) | 0.005* |
| Comorbidity | ||||
| Hypertension | 571 (55.93%) | 346 (57.38%) | 225 (53.83%) | 0.261 |
| Diabetes | 304 (29.77%) | 180 (29.85%) | 124 (29.67%) | 0.949 |
| Congestive heart failure | 346 (33.89%) | 198 (32.84%) | 148 (35.41%) | 0.393 |
| Cerebrovascular disease | 108 (10.58%) | 60 (9.95%) | 48 (11.48%) | 0.434 |
| COPD | 248 (24.29%) | 139 (23.05%) | 109 (26.08%) | 0.268 |
| Renal disease | 259 (25.37%) | 152 (25.21%) | 107 (25.60%) | 0.888 |
| Sepsis | 826 (80.90%) | 459 (76.12%) | 367 (87.80%) | < 0.001* |
| NRBCs (%) | 1.00 (0.30–2.00) | 0.90 (0.20–1.30) | 1.00 (0.30–2.95) | < 0.001* |
Note: * indicates statistical significance.
3.3. Baseline Demographic and Clinical Characteristics in the Low and High NRBC Groups
The 1021 eligible patients were divided into a high NRBC group (≥ 1.0%) and a low NRBC group (< 1.0%) based on NRBC counts. The baseline demographic and clinical characteristics of the patients with liver disease in the two groups are presented in Table 2. Before matching, the two groups showed significant differences in sex; SOFA scores; ALT, AST, Alb and Lac levels; and rates of diabetes, congestive heart failure and sepsis (p < 0.05). After matching, no significant between‐group differences were observed (p > 0.05).
TABLE 2.
The baseline demographic and clinical characteristics of patients with liver disease in the high and low NRBC groups before and after PSM.
| Variable | Before match | After match | ||||||
|---|---|---|---|---|---|---|---|---|
| Low NRBC group (n = 462) (< 1.0%) | High NRBC group (n = 559) (≥ 1.0%) | p | SMD | Low NRBC group (n = 433) (< 1.0%) | High NRBC group (n = 433) (≥ 1.0%) | p | SMD | |
| Demographic data | ||||||||
| Age (year) | 59.02 (14.90) | 58.34 (15.97) | 0.486 | 0.04 | 58.84 (14.93) | 58.67 (15.77) | 0.878 | 0.01 |
| Gender | ||||||||
| Male | 278 (60.17%) | 301 (53.85%) |
0.042 |
0.13 |
252 (58.20%) | 236 (54.50%) | 0.273 | 0.07 |
| Female | 184 (39.83%) | 258 (46.15%) | 181 (41.80%) | 197 (45.50%) | ||||
| Ethnicity | ||||||||
| White | 281 (60.82%) | 326 (58.32%) | 0.719 | 0.05 | 262 (60.51%) | 244 (56.35%) | 0.351 | 0.10 |
| Black | 53 (11.47%) | 68 (12.16%) | 49 (11.32%) | 61 (14.09%) | ||||
| Other | 128 (27.71%) | 165 (29.52%) | 122 (28.18%) | 128 (29.56%) | ||||
| BMI (kg/m2) | 30.75 (6.73) | 30.29 (6.48) | 0.266 | 0.07 | 30.71 (6.88) | 30.54 (6.58) | 0.713 | 0.03 |
| Score system | ||||||||
| Meld score | 25.11 (9.98) | 25.73 (9.80) | 0.322 | 0.06 | 25.18 (9.81) | 25.64 (10.05) | 0.498 | 0.05 |
| SOFA score | 9.00 (6.00–12.00) | 10.00 (7.00–14.00) | < 0.001 | 0.24 | 9.00 (6.00–13.00) | 10.00 (6.00–13.00) | 0.249 | 0.08 |
| AARC score | 10.17 (1.65) | 10.22 (1.69) | 0.641 | 0.03 | 10.19 (1.65) | 10.24 (1.71) | 0.701 | 0.03 |
| Laboratory data | ||||||||
| PLT (K/μL) | 107.00 (65.00–171.00) | 99.00 (56.50–164.50) | 0.236 | 0.07 | 105.00 (64.00–170.00) | 102.00 (60.00–165.00) | 0.406 | 0.06 |
| ALT (IU/L) | 80.00 (27.00–217.00) | 80.00 (37.00–574.00) | < 0.001 | 0.11 | 80.00 (28.00–232.00) | 80.00 (36.00–503.00) | 0.229 | 0.08 |
| AST (IU/L) | 143.00 (54.25–434.25) | 201.00 (84.50–1220.00) | < 0.001 | 0.13 | 148.00 (58.00–474.00) | 172.00 (77.00–1044.00) | 0.287 | 0.07 |
| Alb (g/dL) | 2.87 (0.59) | 2.71 (0.62) | < 0.001 | 0.27 | 2.84 (0.58) | 2.80 (0.61) | 0.353 | 0.06 |
| TBIL (mg/dL) | 2.60 (1.10–6.97) | 2.60 (1.20–7.45) | 0.949 | 0.00 | 2.60 (1.10–7.30) | 2.60 (1.20–7.30) | 0.867 | 0.01 |
| BUN (mg/dL) | 34.00 (19.25–60.00) | 36.00 (21.50–57.50) | 0.923 | 0.01 | 34.00 (19.00–60.00) | 36.00 (21.00–58.00) | 0.940 | 0.01 |
| INR | 1.95 (1.50–2.80) | 1.90 (1.50–2.95) | 0.992 | 0.00 | 2.00 (1.50–2.80) | 1.90 (1.50–3.00) | 0.670 | 0.03 |
| Lac (mmol/L) | 5.76 (3.34) | 6.36 (4.30) | 0.015 | 0.15 | 5.81 (3.42) | 6.00 (3.95) | 0.441 | 0.05 |
| WBC (K/μL) | 15.45 (10.30–21.08) | 14.40 (9.55–22.65) | 0.873 | 0.01 | 15.60 (10.30–21.40) | 14.20 (9.20–22.70) | 0.594 | 0.04 |
| Cr (mg/dL) | 1.80 (1.10–3.30) | 2.10 (1.20–3.40) | 0.804 | 0.02 | 1.80 (1.10–3.30) | 2.10 (1.10–3.50) | 0.868 | 0.01 |
| Comorbidities | ||||||||
| Hypertension | 268 (58.01%) | 303 (54.20%) | 0.223 | 0.08 | 247 (57.04%) | 247 (57.04%) | 1.000 | 0.00 |
| Diabetes | 155 (33.55%) | 149 (26.65%) | 0.016 | 0.15 | 137 (31.64%) | 125 (28.87%) | 0.375 | 0.06 |
| Congestive heart failure | 172 (37.23%) | 174 (31.13%) | 0.040 | 0.13 | 156 (36.03%) | 140 (32.33%) | 0.252 | 0.08 |
| Cerebrovascular disease | 58 (12.55%) | 50 (8.94%) | 0.062 | 0.12 | 47 (10.85%) | 43 (9.93%) | 0.656 | 0.03 |
| COPD | 111 (24.03%) | 137 (24.51%) | 0.858 | 0.01 | 107 (24.71%) | 105 (24.25%) | 0.874 | 0.01 |
| Renal disease | 129 (27.92%) | 130 (23.26%) | 0.088 | 0.11 | 117 (27.02%) | 112 (25.87%) | 0.700 | 0.03 |
| Sepsis | 356 (77.06%) | 470 (84.08%) | 0.004 | 0.18 | 341 (78.75%) | 348 (80.37%) | 0.555 | 0.04 |
3.4. Outcomes of Patients With Liver Disease in NRBC Groups
In the pre‐matched cohort, the 30‐day [46.51% vs. 34.20%], 90‐day [55.28% vs. 43.94%] and in‐hospital mortality [48.12% vs. 35.28%] rates were higher in the high NRBC group than in the low NRBC group. Compared to the low NRBC group, the risk of 30‐day, 90‐day and in‐hospital mortality increased by 51% [hazard ratio (HR): 1.51, 95% confidence interval (CI): 1.24–1.84], 43% (HR: 1.43, 95% CI: 1.19–1.70) and 70% (HR: 1.70, 95% CI: 1.32–2.19), respectively, in the high NRBC group. The LOS‐hospital were 16.71 (8.73–27.12) and 4.63 (2.31–10.76) days, respectively, in the low NRBC group and 14.73 (7.84–25.18) and 5.71 (2.04–12.47) days, respectively, in the high NRBC group.
Similar results were observed in the post‐matching cohort. The 30‐day [45.03% vs. 35.33%], 90‐day [54.50% vs. 44.11%] and in‐hospital [46.19% vs. 36.26%] mortality rates were significantly higher in the high NRBC group than in the low NRBC group. Compared to the low NRBC group, the risk of 30‐day, 90‐day and in‐hospital mortality increased by 37% [HR: 1.37, 95% CI: 1.11–1.70], 36% (HR: 1.36, 95% CI: 1.12–1.64) and 51% (HR: 1.51, 95% CI: 1.15–1.98), respectively, in the high NRBC group. The LOS‐hospital was longer in the low NRBC group [16.36 (8.41–27.23)] than in the high NRBC group [14.51 (8.06–24.75)] (Table 3).
TABLE 3.
Outcomes of the patients with liver disease in the low and high NRBC groups.
| Variables | Low NRBC group | High NRBC group | HR (95% CI) | p |
|---|---|---|---|---|
| (< 1.0%) | (≥ 1.0%) | |||
| Pre‐matched cohort | N = 462 | N = 559 | ||
| 30‐day mortality | 158 (34.20%) | 260 (46.51%) | 1.51 (1.24, 1.84) | < 0.001* |
| 90‐day mortality | 203 (43.94%) | 309 (55.28%) | 1.43 (1.19, 1.70) | < 0.001* |
| In‐hospital mortality | 163 (35.28%) | 269 (48.12%) | 1.70 (1.32, 2.19) | < 0.001* |
| LOS hospital | 16.71 (8.73–27.12) | 14.73 (7.84–25.18) | −2.63 (−5.18, −0.07) | 0.021* |
| LOS ICU | 4.63 (2.31–10.76) | 5.71 (2.04–12.47) | 1.44 (0.36, 2.52) | 0.131 |
| Post‐matched cohort | N = 433 | N = 433 | ||
| 30‐day mortality | 153 (35.33%) | 195 (45.03%) | 1.37 (1.11, 1.70) | 0.004* |
| 90‐day mortality | 191 (44.11%) | 236 (54.50%) | 1.36 (1.12, 1.64) | 0.002* |
| In‐hospital mortality | 157 (36.26%) | 200 (46.19%) | 1.51 (1.15, 1.98) | 0.003* |
| LOS hospital | 16.36 (8.41–27.23) | 14.51 (8.06–24.75) | −2.81 (−5.53, −0.09) | 0.031* |
| LOS ICU | 4.61 (2.32–10.76) | 5.24 (1.94–10.97) | 0.78 (−0.34, 1.90) | 0.643 |
Note: * indicates statistical significance.
3.5. Association Between NRBC Counts and 30‐Day Mortality in the Pre‐Matched Cohort of Patients With Liver Disease
Univariate Cox analysis showed that NRBC counts were positively associated with the 30‐day mortality (HR: 1.03, 95% CI: 1.02–1.04). In addition, age (HR: 1.02, 95% CI: 1.01–1.03), ethnicity (HR: 1.48, 95% CI: 1.20–1.82), MELD score (HR: 1.05, 95% CI: 1.04–1.06), SOFA score (HR: 1.16, 95% CI: 1.13–1.19), AARC score (HR: 1.28, 95% CI: 1.21–1.35), AST level (HR: 1.00, 95% CI: 1.00–1.01), TBIL level (HR: 1.01, 95% CI: 1.00–1.02), BUN level (HR: 1.01, 95% CI: 1.00–1.01), INR (HR: 1.12, 95% CI: 1.09–1.16), Lac level (HR: 1.11, 95% CI: 1.09–1.14), WBC count (HR: 1.02, 95% CI: 1.01–1.03), Cr level (HR: 1.06, 95% CI: 1.02–1.10) and sepsis rate (HR: 1.92, 95% CI: 1.43–2.57) were positively correlated with 30‐day mortality, while PLT counts (HR: 1.00, 95% CI: 1.00–1.00) and Alb levels (HR: 0.68, 95% CI: 0.58–0.81) were negatively correlated (Figure 2).
FIGURE 2.
Variables related to 30‐day mortality.
The results of the multivariate Cox regression analysis are presented in Table 4. A positive association was found between the NRBC count as a continuous variable and 30‐day mortality in the unadjusted model (HR: 1.03, 95% CI: 1.02–1.04), adjusted model I (HR: 1.03, 95% CI: 1.02–1.04) and adjusted model II (HR: 1.02, 95% CI: 1.01–1.03). When the NRBC count was used as a dichotomous categorical variable, compared to the low group (< 1.0%), the risk of 30‐day mortality in the high group (≥ 1.0%) increased by 51%, 51% and 29% in the unadjusted model, model I and model II, respectively. When the NRBC count was categorised into quartiles, the Q4 group (≥ 2.0%) had the highest mortality risk in the unadjusted model (HR: 1.68, 95% CI: 1.29–2.19), compared to the Q1 group (≤ 0.2%). Similar results were observed in adjusted model I (HR: 1.63, 95% CI: 1.25–2.12) and adjusted model II (HR: 1.35, 95% CI: 1.03–1.78). For the sensitivity analysis, the NRBC count quartile categorical variables were considered continuous variables, and the trend in the association between the NRBC count and 30‐day mortality was consistent (p < 0.05).
TABLE 4.
Multivariable cox regression analysis of baseline NRBC for 30‐day mortality in the pre‐matched cohort.
| Exposure | HR (95% CI) | ||
|---|---|---|---|
| Non‐adjusted model | Adjusted model I | Adjusted model II | |
| NRBCs | 1.03 (1.02, 1.04)* | 1.03 (1.02, 1.04)* | 1.02 (1.01, 1.03)* |
| NRBCs dichotomous | |||
| Low (< 1.0%) | Ref | Ref | Ref |
| High (≥ 1.0%) | 1.51 (1.24, 1.84)* | 1.51 (1.24, 1.84)* | 1.29 (1.05, 1.58)* |
| NRBCs quartile | |||
| Q1 (≤ 0.2%) | Ref | Ref | Ref |
| Q2 (0.3%–0.9%) | 0.87 (0.63, 1.19) | 0.81 (0.59, 1.10) | 0.88 (0.64, 1.21) |
| Q3 (1.0%–1.9%) | 1.17 (0.89, 1.55) | 1.14 (0.86, 1.50) | 1.07 (0.80, 1.42) |
| Q4 (≥ 2.0%) | 1.68 (1.29, 2.19)* | 1.63 (1.25, 2.12)* | 1.35 (1.03, 1.78)* |
| p value for trend | < 0.001 | < 0.001 | 0.010 |
Note: Adjusted model I: adjust for age; gender; ethnicity; BMI. Adjusted model II: adjust for age; gender; ethnicity; BMI; Meld; SOFA; ARRC; PLT; AST; Alb; INR; Lac; WBC; TBIL; BUN; Cr; hypertension; sepsis. * indicates statistical significance.
3.6. Nonlinear Relationship Between NRBCs and 30‐Day Mortality Risk Among Patients With Liver Disease
Smoothing splines were used todepict the relationship between NRBCs and 30‐day mortality in individuals with liver disease. Using a segmentation regression model, we identified the inflection point of the NRBC count as 7.0% (Table 5). After adjustment, when the NRBC count was < 7.0%, the mortality risk increased by 8.9% for each unit increment in NRBC count (HR: 1.09; p = 0.001). A p‐value of 0.010 for NRBCs indicated a nonlinear correlation between the NRBC count and 30‐day mortality risk in patients with liver disease (Table 5, Figure 3).
TABLE 5.
Threshold effect analysis.
| Models | Unadjusted | Adjusted | ||
|---|---|---|---|---|
| HR (95% CI) | p | HR (95% CI) | p | |
| Model I | ||||
| One line slope | 1.03 (1.02, 1.04) | < 0.001 | 1.02 (1.01, 1.03) | 0.004 |
| Model II | ||||
| Turning point (K1) | ||||
| < 7.0 slope 1 | 1.15 (1.10, 1.21) | < 0.001 | 1.09 (1.03, 1.15) | 0.001 |
| > 7.0 slope 2 | 1.00 (0.99, 1.02) | 0.862 | 1.00 (0.98, 1.02) | 0.991 |
| LRT | < 0.001 a | 0.010 a | ||
Note: Data are presented as HR (95% CI) and p‐value; LRT, Logarithmic likelihood ratio test (p < 0.05 means model II is significantly different from model I, which indicates a nonlinear association). The adjusted variables included age, gender, ethnicity, BMI, Meld, SOFA, ARRC, PLT, AST, Alb, INR, Lac, WBC, TBIL, BUN, Cr, hypertension and sepsis.
Indicates that model II is significant different from model I. Model I, linear analysis; model II, nonlinear analysis.
FIGURE 3.
Smoothing spline fitting curve.
3.7. Subgroup Analyses of the Effect of NRBCs on 30‐Day Mortality in Patients With Liver Disease
We analysed the effect of NRBCs on 30‐day mortality rates in different liver diseases. NRBCs correlated with an increase in 30‐day mortality among patients with hepatitis (HR: 1.03, 95% CI: 1.02–1.05), cirrhosis (HR: 1.05, 95% CI: 1.02–1.07), other liver diseases (HR: 1.03, 95% CI: 1.02–1.05) and liver failure (HR: 0.99, 95% CI: 0.96–1.03) (Table 6). In addition, logistic regression analysis showed NRBCs were associated with increased mortality among patients with different etiologies, including viral infections (HR: 1.07, 95% CI: 1.04–1.10), alcoholism (HR: 1.03, 95% CI: 1.01–1.04) and other etiologies (HR: 1.02, 95% CI: 1.00–1.03). Additionally, interaction analyses revealed significant interaction terms between disease processes (p = 0.030) and etiologies (p = 0.027) and 30‐day mortality risk.
TABLE 6.
Subgroup analyses of the effect of NRBCs on 30‐day mortality in patients with liver disease.
| Subgroups | N | 30‐day mortality | ||
|---|---|---|---|---|
| HR (95% CI) | p | p for interaction | ||
| Disease process | ||||
| Hepatitis | 369 | 1.03 (1.02, 1.05) | < 0.001* | 0.030* |
| Cirrhosis | 232 | 1.05 (1.02, 1.07) | < 0.001* | |
| Liver failure | 227 | 0.99 (0.96, 1.03) | 0.697 | |
| Other | 193 | 1.03 (1.02, 1.05) | < 0.001* | |
| Aetiology | ||||
| Virus infection | 205 | 1.07 (1.04, 1.10) | < 0.001* | 0.027* |
| Alcoholic | 273 | 1.03 (1.01, 1.04) | < 0.001* | |
| Other | 543 | 1.02 (1.00, 1.03) | 0.028* | |
Note: * indicates statistical significance.
3.8. Correlation of NRBC Count Changes With 30‐Day Mortality in Patients With Liver Disease
The changes in NRBC counts over time between the two groups are shown in Figure 4 and Table 7. In the 30‐day survivor group, NRBC counts decreased over time, whereas in the 30‐day non‐survivor group, NRBC counts first declined and then continued to increase (Figure 4). The 30‐day non‐survivors experienced an average increase of 0.31% per day in NRBC counts compared to the 30‐day survivors (Table 7).
FIGURE 4.
Association between changes in NRBCs and 30‐day mortality.
TABLE 7.
Relationship between changes (0–30 days) NRBC counts and 30‐day mortality in patients with liver disease derived from a generalised additive mixed model (GAMM).
| Models | β (95% CI) | p |
|---|---|---|
| Day | −0.08 (−0.20, 0.04) | 0.184 |
| Death | 0.89 (−1.11, 2.90) | 0.384 |
| Day × death | 0.31 (0.12, 0.50) | 0.002* |
Note: Day indicated the mean of the increasing of indexes daily in 30‐day survival patients. Death indicated the difference of indexes at day 0 in 30‐day non‐survival patients compared with 30‐day survival patients. Day × Death indicated the average increasing of indexes daily in 30‐day non‐survival patients compared with 30‐day survival patients. * indicates statistical significance.
4. Discussion
NRBCs are typically absent from the peripheral blood of healthy adults. While previous studies have indicated that NRBCs in peripheral blood are linked to a poorer prognosis in various diseases, their clinical importance in liver diseases remains uncertain. This study explored the potential of NRBCs as predictors of mortality in a diverse group of patients with liver disease using the MIMIC‐IV (v2.0) database. First, we found that compared with the 30‐day survival group, the 30‐day mortality group had higher NRBC counts. All patients were categorised into low‐ and high‐NRBC groups based on their NRBC counts. High NRBC counts were linked to increasing 30‐day, 90‐day and in‐hospital mortalities but reduced LOS‐hospital. This may be due to the high mortality rate of patients in the high NRBC count group, resulting in a corresponding reduction in the average LOS‐hospital. Tayyab et al. [27] reported similar findings. The authors found that ICU patients with NRBCs had significantly higher mortality rates than those without NRBCs. Sarah et al. [28] observed that NRBCs were associated with notably higher mortality in neonatal ICU patients. These results reinforce the significant association between NRBCs and increased disease‐related mortality.
Next, after adjusting for different covariates, we analysed the relationship between NRBCs and mortality in different models. Multivariate Cox regression analysis revealed that the association between the NRBC count and mortality was consistent among the different models. In addition, the mortality risk increased with NRBC counts < 7.0 but not with NRBC counts > 7.0. Further subgroup analyses of the effect of NRBCs on 30‐day mortality in patients with different liver diseases indicated that NRBCs were associated with higher 30‐day mortality in patients with hepatitis, cirrhosis and other liver disease processes in addition to liver failure. Krieg et al. [29] reported similar findings inpatients receiving surgical intensive care. They found that the presence of NRBCs in the blood of patients in surgical intensive care has prognostic significance for patient mortality but is not linked to liver or kidney organ failure [29]. Lastly, from a GAMM, we found that 30‐day non‐survivors experienced an average increase in NRBC counts of 0.31% per day compared to 30‐day survivors, suggesting that NRBCs could be a useful tool for identifying critically ill patients with liver disease at a high risk of death. Furthermore, we found that patients with high NRBC counts had an average daily increase in the TBIL level (0.16 mg/dL), DBIL level (0.26 mg/dL), IBIL level (0.05 mg/dL), Cre level (0.01 mg/dL), INR (0.01) and WBC count (0.09 K/μL) compared with patients with low NRBC counts (Figure S1, Table S1), suggesting that NRBCs may affect the mortality of patients with liver disease by aggravating liver and kidney injuries.
However, the specific biological mechanisms linking NRBCs to the onset and progression of liver disease and mortality remain unclear. These potential pathways may be involved in immune regulation and responses. As the precursors of mature red blood cells, NRBCs have a vital but frequently neglected function in the immunological landscape of disorders [30, 31]. At their nucleation stage, NRBCs possess unique properties that intertwine with immune regulation and responses. Immunologically, NRBCs express both CD71 and CD235a in humans and CD71 and TER119 in mice, known as CD71+ erythroid cells (CECs), which include NRBCs and reticulocytes [32, 33]. CECs influence the host defence system in several ways. They have been implicated in modulating inflammatory responses through the secretion of cytokines and the expression of pattern recognition receptors, such as arginase‐2, TGF‐β, galectins, PD‐1: PDL‐1/PDL‐2, VISTA and ROS, which enable them to participate in pathogen detection and phagocytosis [31, 34, 35, 36]. Moreover, CECs can present antigens to T cells, thereby contributing to adaptive immunity [37]. The immunosuppressive and immunomodulatory properties of CECs have been observed in several diseases. In human immunodeficiency virus (HIV)‐infected patients, the CEC frequency is positively related to the plasma viral load. In addition, HIV particles capable of infection are located inside CECs but are absent in mature red blood cells. More significantly, CECs, through ROS, enhance HIV replication and infection in CD4+ T cells, making them more prone to HIV infection [38]. In patients with sepsis, CECs are expanded in the peripheral blood, and their frequencies are positively correlated with high levels of interleukin (IL)‐6 and interferon (IFN)‐γ. Furthermore, CECs may act as standalone indicators of the onset of hospital‐acquired infections and mortality within 30 days [39]. All these studies emphasise the crucial function of CECs in immunomodulation. Our findings indicate that NRBCs are crucial for the progression of liver diseases. On this basis, additional research is required to ascertain whether better management of NRBCs will enhance clinical outcomes, how NRBCs function in liver diseases (e.g., by modulating the immune response), and whether NRBCs could serve as a new target for improving prognoses and clinical outcomes in this population.
Research on liver disease biomarkers presents both promises and challenges in disease diagnosis and prognostic evaluation. The specificities and sensitivities of these biomarkers are critical. Currently, the diagnostic accuracy of many identified biomarkers requires further improvement because false‐positive or false‐negative results limit their clinical application. For example, alpha‐fetoprotein (AFP), a commonly used biomarker for HCC, may not be elevated in some patients with HCC and can yield false‐positive results in certain benign liver diseases, thereby compromising diagnostic reliability [40]. Additionally, variations in detection methods and standards across studies hinder the comparability and validation of results, further complicating the clinical application of biomarkers. Research on liver disease biomarkers offers numerous opportunities. With continuous technological advancements, the development of multiomics technologies, such as genomics, transcriptomics, proteomics and metabolomics, provides powerful tools for discovering additional potential biomarkers. By integrating and analysing biological information across different levels, it is possible to identify biomarker combinations with higher specificity and sensitivity, thereby improving the accuracy of disease diagnosis and prognosis evaluation. Furthermore, the rise of interdisciplinary research has fostered collaboration across fields such as medicine, biology and engineering, offering new perspectives and methodologies for biomarker research and applications. For instance, advances in nanotechnology and biosensors have enabled more sensitive and convenient detection methods for biomarkers. In light of these challenges and opportunities, collaborative efforts among researchers, clinicians and policymakers are essential to advance the study of liver disease biomarkers and improve patient health outcomes.
The major strength of this study was the evidence that elevated NRBC counts independently contribute to a higher mortality risk in patients with liver disease. In addition, dynamic changes in NRBC counts were available. Therefore, we assessed the predictive ability of changes in NRBC counts. However, this study has several limitations. First, residual confounding factors may have affected the clinical outcomes despite the use of multivariate adjustment and subgroup analyses; therefore, additional prospective case–control studies are needed. Secondly, this study did not explore the mechanism by which NRBCs affect liver diseases, and more experiments are needed to delve deeper into the mechanisms involved.
5. Conclusion
In summary, our results indicated that elevated NRBC counts are correlated with an increased risk of 30‐day, 90‐day and in‐hospital mortality in critically ill patients with liver diseases, confirming the significant link between NRBCs in the peripheral blood and higher mortality rates from diseases. Monitoring NRBC counts may aid in decision‐making and the management of liver diseases in clinical settings.
Author Contributions
Y.Z.: Writing – original draft; D.Y.: Writing – review and editing; F.P.: Formal analysis; R.Y.: Software; N.L.: Funding acquisition. All authors read and approved the final manuscript.
Funding
This work was funded by the National Natural Science Foundation of China (No. 82571796) and the Clinical Research Center Foundation of Xiangya Hospital (No. LN2021XYSX).
Ethics Statement
The study employed public data sourced from the MIMIC‐IV database, with the data being de‐identified and not altered or combined in any way that would jeopardise participant confidentiality. Thus, the research does not need any more approval or extra consent.
Consent
The authors have nothing to report.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Data S1: jcmm70982‐sup‐0001‐DataS1.docx.
Acknowledgements
The authors have nothing to report.
Zhu Y., Yan D., Peng F., Yao R., and Li N., “Association Between Nucleated Red Blood Cell Counts and the Mortality in Patients With Liver Diseases: An Analysis of the MIMIC‐IV Database,” Journal of Cellular and Molecular Medicine 29, no. 23 (2025): e70982, 10.1111/jcmm.70982.
Yuanyuan Zhu and Danyang Yan contributed equally.
Contributor Information
Run Yao, Email: yaorunxy@csu.edu.cn.
Ning Li, Email: liningxy@csu.edu.cn.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
- 1. Trefts E., Gannon M., and Wasserman D. H., “The Liver,” Current Biology 27, no. 21 (2017): R1147–R1151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Hora S. and Wuestefeld T., “Liver Injury and Regeneration: Current Understanding, New Approaches, and Future Perspectives,” Cells 12, no. 17 (2023): 2129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Cichoż‐Lach H. and Michalak A., “Oxidative Stress as a Crucial Factor in Liver Diseases,” World Journal of Gastroenterology 20, no. 25 (2014): 8082–8091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Hammerich L. and Tacke F., “Hepatic Inflammatory Responses in Liver Fibrosis,” Nature Reviews. Gastroenterology & Hepatology 20, no. 10 (2023): 633–646. [DOI] [PubMed] [Google Scholar]
- 5. Hassan G. S., Flores Molina M., and Shoukry N. H., “The Multifaceted Role of Macrophages During Acute Liver Injury,” Frontiers in Immunology 14 (2023): 1237042. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Hsu C. L. and Schnabl B., “The Gut‐Liver Axis and Gut Microbiota in Health and Liver Disease,” Nature Reviews. Microbiology 21, no. 11 (2023): 719–733. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. McConnell M. J., Kostallari E., Ibrahim S. H., and Iwakiri Y., “The Evolving Role of Liver Sinusoidal Endothelial Cells in Liver Health and Disease,” Hepatology 78, no. 2 (2023): 649–669. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Khan S., Mahgoub S., Fallatah N., Lalor P. F., and Newsome P. N., “Liver Disease and Cell Therapy: Advances Made and Remaining Challenges,” Stem Cells 41, no. 8 (2023): 739–761. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Devarbhavi H., Asrani S. K., Arab J. P., Nartey Y. A., Pose E., and Kamath P. S., “Global Burden of Liver Disease: 2023 Update,” Journal of Hepatology 79, no. 2 (2023): 516–537. [DOI] [PubMed] [Google Scholar]
- 10. Palis J. and Segel G. B., “Developmental Biology of Erythropoiesis,” Blood Reviews 12, no. 2 (1998): 106–114. [DOI] [PubMed] [Google Scholar]
- 11. Dzierzak E. and Philipsen S., “Erythropoiesis: Development and Differentiation,” Cold Spring Harbor Perspectives in Medicine 3, no. 4 (2013): a011601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Menk M., Giebelhäuser L., Vorderwülbecke G., et al., “Nucleated Red Blood Cells as Predictors of Mortality in Patients With Acute Respiratory Distress Syndrome (ARDS): An Observational Study,” Annals of Intensive Care 8, no. 1 (2018): 42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Kuert S., Holland‐Letz T., Friese J., and Stachon A., “Association of Nucleated Red Blood Cells in Blood and Arterial Oxygen Partial Tension,” Clinical Chemistry and Laboratory Medicine 49, no. 2 (2011): 257–263. [DOI] [PubMed] [Google Scholar]
- 14. Stachon A., Segbers E., Holland‐Letz T., Kempf R., Hering S., and Krieg M., “Nucleated Red Blood Cells in the Blood of Medical Intensive Care Patients Indicate Increased Mortality Risk: A Prospective Cohort Study,” Critical Care 11, no. 3 (2007): R62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Narcı H., Oktay M. M., Ayrık C., and Çimen M. B. Y., “Nucleated Red Blood Cells as Predictor of All‐Cause Mortality in Emergency Department,” American Journal of Emergency Medicine 46 (2021): 335–338. [DOI] [PubMed] [Google Scholar]
- 16. Pikora K., Krętowska‐Grunwald A., Krawczuk‐Rybak M., and Sawicka‐Żukowska M., “Diagnostic Value and Prognostic Significance of Nucleated Red Blood Cells (NRBCs) in Selected Medical Conditions,” Cells 12, no. 14 (2023): 1817. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Li H., Tu Q., Feng K., et al., “Nucleated Red Blood Cells as a Novel Biomarker in the Diagnosis and Prediction of Sepsis Severity in Children,” Frontiers in Cellular and Infection Microbiology 13 (2023): 1264607. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Desai S., Jones S. L., Turner K. L., Hall J., and Moore L. J., “Nucleated Red Blood Cells Are Associated With a Higher Mortality Rate in Patients With Surgical Sepsis,” Surgical Infections 13, no. 6 (2012): 360–365. [DOI] [PubMed] [Google Scholar]
- 19. Monteiro Júnior J. G., Torres D. O., da Silva M. C., et al., “Nucleated Red Blood Cells as Predictors of All‐Cause Mortality in Cardiac Intensive Care Unit Patients: A Prospective Cohort Study,” PLoS One 10, no. 12 (2015): e0144259. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Yang L., Shivakumar P., Kinder J., et al., “Regulation of Bile Duct Epithelial Injury by Hepatic CD71+ Erythroid Cells,” JCI Insight 5, no. 11 (2020): e135751. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Chen J., Qiao Y. D., Li X., et al., “Intratumoral CD45(+) CD71(+) Erythroid Cells Induce Immune Tolerance and Predict Tumor Recurrence in Hepatocellular Carcinoma,” Cancer Letters 499 (2021): 85–98. [DOI] [PubMed] [Google Scholar]
- 22. Kim W. R., Mannalithara A., Heimbach J. K., et al., “MELD 3.0: The Model for End‐Stage Liver Disease Updated for the Modern Era,” Gastroenterology 161, no. 6 (2021): 1887–1895. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Raith E. P., Udy A. A., Bailey M., et al., “Prognostic Accuracy of the SOFA Score, SIRS Criteria, and qSOFA Score for In‐Hospital Mortality Among Adults With Suspected Infection Admitted to the Intensive Care Unit,” Journal of the American Medical Association 317, no. 3 (2017): 290–300. [DOI] [PubMed] [Google Scholar]
- 24. Maiwall R., Pasupuleti S. S. R., Choudhury A., et al., “AARC Score Determines Outcomes in Patients With Alcohol‐Associated Hepatitis: A Multinational Study,” Hepatology International 17, no. 3 (2023): 662–675. [DOI] [PubMed] [Google Scholar]
- 25. Loke Y. K. and Mattishent K., “Propensity Score Methods in Real‐World Epidemiology: A Practical Guide for First‐Time Users,” Diabetes, Obesity & Metabolism 22, no. Suppl 3 (2020): 13–20. [DOI] [PubMed] [Google Scholar]
- 26. Jaddoe V. W., de Jonge L. L., Hofman A., Franco O. H., Steegers E. A., and Gaillard R., “First Trimester Fetal Growth Restriction and Cardiovascular Risk Factors in School Age Children: Population Based Cohort Study,” British Medical Journal 348 (2014): g14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Noor T., Imran A., Raza H., Sarwar M., Umer S., and Fatima M., “Frequency of Nucleated Red Blood Cells in the Peripheral Blood of ICU‐Admitted Patients,” Cureus 15, no. 1 (2023): e33827. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Morton S. U., Brettin K., Feldman H. A., and Leeman K. T., “Association of Nucleated Red Blood Cell Count With Mortality Among Neonatal Intensive Care Unit Patients,” Pediatrics and Neonatology 61, no. 6 (2020): 592–597. [DOI] [PubMed] [Google Scholar]
- 29. Stachon A., Holland‐Letz T., Kempf R., Becker A., Friese J., and Krieg M., “Poor Prognosis Indicated by Nucleated Red Blood Cells in Peripheral Blood Is Not Associated With Organ Failure of the Liver or Kidney,” Clinical Chemistry and Laboratory Medicine 44, no. 8 (2006): 955–961. [DOI] [PubMed] [Google Scholar]
- 30. Elahi S., Ertelt J. M., Kinder J. M., et al., “Immunosuppressive CD71+ Erythroid Cells Compromise Neonatal Host Defence Against Infection,” Nature 504, no. 7478 (2013): 158–162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Shahbaz S., Bozorgmehr N., Koleva P., et al., “CD71+VISTA+ Erythroid Cells Promote the Development and Function of Regulatory T Cells Through TGF‐β,” PLoS Biology 16, no. 12 (2018): e2006649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Grzywa T. M., Nowis D., and Golab J., “The Role of CD71(+) Erythroid Cells in the Regulation of the Immune Response,” Pharmacology & Therapeutics 228 (2021): 107927. [DOI] [PubMed] [Google Scholar]
- 33. Chen K., Liu J., Heck S., Chasis J. A., An X., and Mohandas N., “Resolving the Distinct Stages in Erythroid Differentiation Based on Dynamic Changes in Membrane Protein Expression During Erythropoiesis,” Proceedings of the National Academy of Sciences of the United States of America 106, no. 41 (2009): 17413–17418. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Seledtsova G. V., Seledtsov V. I., Samarin D. M., et al., “Erythroid Cells in Immunoregulation: Characterization of a Novel Suppressor Factor,” Immunology Letters 93, no. 2–3 (2004): 171–178. [DOI] [PubMed] [Google Scholar]
- 35. Delyea C., Bozorgmehr N., Koleva P., et al., “CD71(+) Erythroid Suppressor Cells Promote Fetomaternal Tolerance Through Arginase‐2 and PDL‐1,” Journal of Immunology 200, no. 12 (2018): 4044–4058. [DOI] [PubMed] [Google Scholar]
- 36. Zhao L., He R., Long H., et al., “Late‐Stage Tumors Induce Anemia and Immunosuppressive Extramedullary Erythroid Progenitor Cells,” Nature Medicine 24, no. 10 (2018): 1536–1544. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Sanjabi S., Oh S. A., and Li M. O., “Regulation of the Immune Response by TGF‐β: From Conception to Autoimmunity and Infection,” Cold Spring Harbor Perspectives in Biology 9, no. 6 (2017): a022236. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Namdar A., Dunsmore G., Shahbaz S., et al., “CD71(+) Erythroid Cells Exacerbate HIV‐1 Susceptibility, Mediate Trans‐Infection, and Harbor Infective Viral Particles,” MBio 10, no. 6 (2019): e02767‐19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Zhao G. J., Jiang D. W., Cai W. C., et al., “CD71(+) Erythroid Cell Expansion in Adult Sepsis: Potential Causes and Role in Prognosis and Nosocomial Infection Prediction,” Frontiers in Immunology 13 (2022): 830025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Stefanini B., Tonnini M., Serio I., Renzulli M., and Tovoli F., “Surveillance for Hepatocellular Carcinoma: Current Status and Future Perspectives for Improvement,” Expert Review of Anticancer Therapy 22, no. 4 (2022): 371–381. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1: jcmm70982‐sup‐0001‐DataS1.docx.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.