Sepsis has an in-hospital mortality of 25.8%, and has been one of the leading causes of death in critically ill patients.^[1,2] Inflammation-associated anemia is mainly due to the involvement of inflammatory mediators, which is commonly observed in patients with sepsis. It often develops in patients with long-term infection, inflammatory disease, or malignancy, accompanied by low serum iron despite adequate systemic iron store, decreased serum transferrin, and mild decrease in size and hemoglobin content of erythrocytes.^[3] Interleukin (IL)-6 is a key inflammatory cytokine inducing an abrupt increase in hepcidin synthesis, thereby decreasing the serum iron in sepsis patients.^[3] The survival of sepsis patients is closely associated with parameters that reflect inflammation-related iron metabolism disorders and anemia. Jiang et al^[3] have observed significantly decreased serum iron, hemoglobin, and soluble transferrin receptor (sTfR)/log ferritin, with remarkably increased serum erythropoietin, hepcidin, ferritin, and IL-6 in sepsis patients during the first week of intensive care unit (ICU) admission. More importantly, the 28-day mortality was closely associated with the changes in serum hepcidin, ferritin, and IL-6.^[3]

At present, effective treatment for severe inflammation-related anemia and iron metabolism disorders in sepsis patients remains unclear. Moreover, conventional blood transfusion and iron supplement are not routinely recommended in the current clinical practice. Blood transfusion has been reported to be associated with higher mortality and failure of anemia correction in critically ill patients.^[4] Although enteral iron supplementation can reduce the blood transfusion rate in critically ill patients with baseline iron deficiency, iron remains as a toxic compound that may induce oxidative stress and bacterial growth.^[5]

Continuous renal replacement therapy (CRRT) has been traditionally used as an important treatment strategy for renal failure.^[6,7] CRRT can non-selectively remove inflammatory mediators, rebalance acid-base disorders, regulate immune stability, and maintain a stable environment through the mechanisms of convection and ultrafiltration.^[8] Therefore, CRRT is applicable for eliminating inflammatory mediators in sepsis.^[9] However, the effects of CRRT on inflammation-related anemia and iron metabolism disorders in sepsis remain unclear. The present prospective study aimed to determine whether CRRT would affect the parameters of inflammation-related anemia, iron metabolism, and prognosis in sepsis patients with acute kidney injury (AKI).

Sepsis patients with AKI, who were admitted to the emergency ICU at the First Affiliated Hospital of Dalian Medical University from October 2015 to December 2017, were prospectively enrolled. All patients met the Sepsis-2 (severe sepsis between October 2015 and February 2016) or Sepsis-3 (sepsis after the publishment of the Sepsis-3 criteria on February 23, 2016) criteria, and had sepsis-induced AKI with rapid increase in creatinine, or oliguria with fluid overload and/or hyperkalemia.^[6,7] AKI was diagnosed according to the Kidney Disease Improving Global Outcomes (KDIGO) guidelines.^[10]

The exclusion criteria were as follows: <18 years old; pregnant patients; patients with stomach cancer and acute bleeding; patients with blood system diseases (e.g., leukemia, myeloma, red blood cell diseases including polycythemia vera, abnormal mean red blood cell volume, abnormal mean hemoglobin content, or abnormal mean hemoglobin concentration); patients with chronic renal insufficiency requiring hemodialysis; patients with liver cirrhosis; patients with prior blood transfusion within one week before admission or during hospitalization; patients with anemia, who were treated with iron and erythropoietin within three months before admission.

All enrolled sepsis patients were randomized into the CRRT (patients who received CRRT) and control (patients who received conventional treatment without CRRT) groups using computer-generated, pre-randomized sealed envelopes. The patients and their caregivers were not blinded to the allocation of treatment.

The present study was performed in accordance with the Declaration of Helsinki adopted by the World Medical Association (2013 edition). The study protocol was approved by the Medical Ethics Committee of the First Affiliated Hospital of Dalian Medical University (PJ-KS-KY-2020-138). A written informed consent was obtained from each participant or legal guardian on admission. The present study was registered to the Chinese Clinical Trial Registry (ChiCTR2100048817, http://www.chictr.org.cn), and followed the Consolidated Standards of Reporting Trials (CONSORT) guidelines.

Patients in the control group received conventional treatment according to the Surviving Sepsis Campaign guidelines.^[11] The patients whose renal function deteriorated also received delayed CRRT, and they were included in the CRRT group. Patients in the CRRT group received CRRT within 24 h after the diagnosis of sepsis-induced AKI, in addition to conventional treatment. The CRRT protocol was provided in the supplementary Figure 1.

The following clinical data were collected on admission: gender, age, infection sites, comorbidities, number of post-operative patients, number of patients who used vasopressors, number of patients who received mechanical ventilation, laboratory results, and 28-day mortality. Furthermore, the hemoglobin, red blood cell distribution width (RDW), reticulocytes, mean corpuscular volume, mean corpuscular hemoglobin, and mean corpuscular hemoglobin concentration were measured using an automatic blood cell analyzer (XN-2000; Sysmex, Japan). The Sequential Organ Failure Assessment (SOFA) score was assessed on ICU admission.

Peripheral venous blood (5 mL) was drawn on days 1 (before CRRT), 3 and 7 of ICU admission. Then, the collected blood was centrifuged (TGL-20MS [BIORIDGE, China], 4 °C, 2,000×g) for 10 min, and stored at -80 °C for the subsequent analysis. Afterwards, the serum IL-6, hepcidin (Uscn Life Sciences, China), erythropoietin, ferritin (Abcam, Cambridge, USA), and sTfR (R&D Systems, USA) were detected by enzyme-linked immunosorbent assay (ELISA).

The data were analyzed using SPSS 23.0 (IBM, USA). Continuous variables were expressed as mean±standard deviation (SD) for normal distribution, or median (interquartile range) for skewed distribution. The categorical variables were compared using Pearson’s Chi-square test or Fisher’s exact test. Continuous variables were compared using Mann-Whitney U-test or t-test. A P-value of <0.05 was considered statistically significant.

A total of 179 consecutive sepsis patients with AKI were initially enrolled for the present study. Among these patients, 99 patients were randomized into the CRRT (n=49) and control (n=50) groups and 80 patients were excluded from the study (supplementary Figure 2). After randomization, 10 patients in the CRRT group abandoned the therapy due to financial problems or other causes. Finally, 39 patients in the CRRT group and 50 patients in the control group were included for the present study. The number (15 vs. 74) and proportion (16.9% vs. 83.1%) of included patients meeting the criterion for Sepsis-2 were lower due to the short study duration (4 months), when compared with those meeting the criterion for Sepsis-3 with a long study duration (22 months). The proportion (16.7% vs. 17.1%) of patients in the control group who met the criterion of Sepsis-2 was also similar to that of patients in the CRRT group who met the criterion of Sepsis-3 (P>0.05). Moreover, the proportion of patients with AKI stage 1, 2 and 3 did not significantly differ between the control (32/50, 18/50, and 0/50, respectively) and CRRT (24/39, 15/39, and 0/39, respectively) groups (P>0.05). Two patients in the stage 2 in the control group, who received delayed CRRT within 24 h of ICU admission due to renal function deterioration, were included in the CRRT group, but both of them died.

There were no significant differences in baseline characteristics, including age, male gender, infection sites, comorbidities, number of postoperative patients, number of patients who used vasopressors, number of patients who received mechanical ventilation, length of mechanical ventilation, PaO₂/FiO₂, procalcitonin, lactate, creatinine, urine output, potassium, reticulocytes, mean corpuscular volume, mean corpuscular hemoglobin, mean corpuscular hemoglobin concentration, or SOFA score, between the two groups (Table 1 and supplementary Table 2). The parameters for the CRRT are summarized in supplementary Table 1. The median survival for hemofiltration was 33.5 h, which was determined by early recovery of renal function or failure to maintain the function of CRRT. The latter was mainly due to elevated transmembrane pressure in three patients with secondary sepsis after surgery, who received no heparinization during CRRT. None of the patients in the CRRT group suffered from bleeding due to anticoagulant use.

There were no significant differences in hemoglobin, RDW, and serum erythropoietin on admission between the two groups (all P>0.05). Furthermore, the hemoglobin and erythropoietin gradually decreased in both groups during the first week (all P<0.05; Figures 1A and 1C), but these did not significantly differ between the two groups. Moreover, the RDW was higher in both groups on days 3 and 7, when compared to that on day 1, but there was no significant difference in RDW between days 3 and 7 (both P>0.05; Figure 1B). In addition, on days 3 and 7, the RDW was significantly lower in the CRRT group, when compared to the control group (both P<0.05).

On admission, the serum IL-6 did not significantly differ between the two groups (P>0.05, Figure 1D). Furthermore, a gradual decrease in serum IL-6 was observed during the first week in both groups (all P<0.05). However, on days 3 and 7, the serum IL-6 were lower in the CRRT group, when compared to the control group (both P<0.05).

There were no significant differences in serum iron, hepcidin, ferritin or sTfR on admission between the two groups (all P>0.05; Figure 2). The serum iron in the control group initially slightly decreased, and subsequently slightly increased in the first week (all P>0.05; Figure 2A), while the serum iron in the CRRT group was higher on day 7, when compared to that on day 1. In addition, on day 7, the serum iron was higher in the CRRT group, when compared to the control group (P<0.05). Furthermore, the serum hepcidin in both groups were lower on days 3 and 7, when compared to those on day 1 (all P<0.05; Figure 2(B)). On days 3 and 7, the serum hepcidin was lower in the CRRT group, when compared to the control group (both P<0.05). However, the serum ferritin in both groups was higher on days 3 and 7, when compared to that on day 1 (all P<0.05; Figure 2(C)), while on days 3 and 7, the serum ferritin was significantly lower in the CRRT group, when compared to the control group (both P<0.05). There were no significant differences in serum sTfR between the two groups, or among the different time points (all P>0.05; Figure 2D).

The serum creatinine and SOFA scores in both groups gradually decreased from day 1 to day 7 (all P<0.05). However, patients in the CRRT group had lower serum creatinine levels and SOFA scores on day 7 after admission to the ICU, improved hemodynamic parameters on days 3 and 7 after admission to the ICU, and shorter ICU stays, when compared to patients in the control group (all P<0.05; supplementary Table 2). There was no significant difference in 28-day mortality (38.0% vs. 28.2%) (supplementary Table 2) between the control and CRRT groups.

The present study evaluating the effect of CRRT on inflammation-related anemia, iron metabolism, and the prognosis of sepsis patients with AKI had three main findings. First, the CRRT did not significantly alter the hemoglobin, erythropoietin or sTfR levels in sepsis patients with AKI. Second, patients with sepsis had significantly higher serum levels of IL-6, hepcidin, ferritin and RDW, but these parameters decreased after CRRT. Third, CRRT was associated with the increase in serum iron and improvement in SOFA scores but not the 28-day mortality.

The incidence of sepsis-related anemia reached as high as 90% within three days after ICU admission. The anemia was mainly caused by a variety of mechanisms, including infection, iatrogenic blood loss, dilution during fluid resuscitation, decrease in serum iron, inhibition of erythropoietin production, shortened red blood cell life, and malnutrition.^[12] For sepsis patients, inflammation upregulates the generation of hepcidin which in turn inhibit serum iron, but these changes exert a measurable effect on hemoglobin levels only from seven days after the diagnosis of sepsis.^[13] Therefore, the effect of decreased hepcidin by CRRT on hemoglobin production may be insignificant.

RDW is typically used as a part of the complete blood count to quantify size changes of circulating red blood cells.^[14] For patients with sepsis, a significant increase in RDW was observed on admission, which may serve as a useful predictor of mortality as evidenced by a previous study.^[3] In the present study, we found that sepsis patients had lower RDWs following CRRT. However, the mechanism by which CRRT affects RDW remains unknown. Furthermore, it was observed that RDW was strongly correlated with IL-6.^[3,14] Moreover, pro-inflammation cytokines, such as IL-6, can impair the maturation of red blood cells, thereby accelerating the entrance of immature red blood cells into circulation.^[14] Therefore, we speculated that the decrease in RDW after CRRT, at least in part, may be correlated to the clearance of IL-6.

It was previously identified that serum erythropoietin is elevated with a tendency to gradually decrease in patients with sepsis.^[3] In the present study, no significant change in serum erythropoietin levels was observed in sepsis patients after CRRT. This phenomenon could be explained by the fact that erythropoietin has a molecular weight of 34 kDa, which makes it difficult to be filtered by the membrane during CRRT.^[15] In addition, the effects of CRRT on erythropoietin production are complicated. Erythropoietin production is usually elevated in response to anemia and hypoxemia, and this is inhibited by impaired renal function and elevated pro-inflammatory cytokines (such as IL-6 and TNF-α) in sepsis patients.^[3] Furthermore, CRRT may ameliorate the hypoxemia and acute lung injury, thereby decreasing erythropoietin production.^[16] Meanwhile, CRRT may remove pro-inflammatory cytokines, thereby increasing erythropoietin production. Considering the inconsistent effects of CRRT on erythropoietin production, it was not surprising to observe that the CRRT failed to affect the serum erythropoietin in sepsis patients in the present study.

Serum IL-6, hepcidin, iron, ferritin and sTfR are parameters of inflammation-related iron metabolism. A previous study found that these parameters, except for serum iron, significantly increased in patients with sepsis, and that serum IL-6 was positively correlated with serum hepcidin and ferritin.^[3] Hepcidin, as a liver-producing peptide hormone, plays a central role in regulating serum iron.^[13] This inhibits iron absorption from the duodenum, and its release from hepatocytes and circulating macrophages, resulting in a low level of serum iron.^[17] IL-6 can induce the upregulation of hepcidin through the STAT-3 signaling pathway, thereby causing alterations in iron metabolism. These alterations were observed to be significantly associated with anemia and 28-day mortality.^[3,5,18] Notably, the present study revealed that CRRT significantly reduced the serum IL-6, hepcidin and ferritin, rather than the serum sTfR.

The decrease in serum IL-6 after CRRT was consistent with that in previous findings, in which inflammatory mediators, such as serum IL-6 and TNF-α, can be cleared by hemofiltration in animal and human experiments.^[6,8,11] The mechanism by which CRRT clears the serum IL-6 may be the adsorption, rather than the convection, because IL-6 has a molecular weight of 26 kDa, making it difficult to be filtered by membrane during CRRT.^[19] Furthermore, CRRT can effectively adsorb lipopolysaccharide, which is an activator of cytokines production. The rapid removal of lipopolysaccharide during CRRT can lead to the decrease in production of IL-6.^[19] Similarly, hepcidin, with a molecular weight of approximately 10 kDa, can be effectively removed through the membrane during CRRT.^[20] In addition, the reduction in serum hepcidin and ferritin in sepsis patients after CRRT may be associated with the reduction in serum IL-6.^[18] This is because the clearance of IL-6 during CRRT can relieve the upregulation of hepcidin through the IL-6-hepcidin axis, and block the induction of pro-inflammatory cytokines to ferritin production.^[18] Considering that hepcidin plays a vital role in decreasing serum iron,^[13] it was speculated that the elevated serum iron after CRRT may be mainly associated with the decrease in serum hepcidin. The serum sTfR levels appeared to be unaffected by the CRRT in sepsis patients in the present study. This was mainly because sTfR has a large molecular weight of 85 kDa, resulting in a very low clearance rate during CRRT. Furthermore, sTfR reflects the degree of iron availability for cells, but this is not affected by inflammation. Therefore, the clearance of IL-6 during CRRT does not affect sTfR production.

Undoubtedly, the correction of iron metabolism disorders may only partially contribute to the improvement in SOFA scores after CRRT in the present study. Increasing evidence has revealed that CRRT can exert multiple effects on sepsis patients, including the correction of acidemia, electrolyte imbalance and extracellular volume expansion, the removal of inflammatory mediators, and the improvement of AKI, and these are all associated with the improvement in SOFA scores.^[6,7,9,18] However, the 28-day mortality of sepsis patients did not improve after CRRT in the present study, which is inconsistent with the meta-analysis results published by Putzu et al.^[7] There are several explanations for this result. First, the relatively small sample size of the present study may not be adequate to reach a statistically significant result. Second, iron metabolism disorders only partially contribute to the severity of sepsis, and these may be insufficient to affect the outcome. Other complications and comorbidities may also influence the mortality.

There were several limitations in the present study. First, the differences in the subgroups of gender and septic shock were not assessed. However, the gender distribution and number of patients who used vasopressors on admission were similar between the two groups. Second, the pre-dilution might have reduced the removal of target cytokines. Third, the replacement of extracorporeal circuits every 48 h was not a sound strategy, but the efficacy of the AN69ST membrane might have been reduced due to the prolonged use. Fourth, the loss rate of follow-up was close to 20% in the CRRT group, which might introduce a bias. Fifth, the current study did not involve the exploration of potential mechanisms. Finally, the sample size was relatively small, and subgroups of septic patients with different etiologies such as urinary tract infection were not all included. Hence, future studies with large sample sizes are warranted.

CRRT might have beneficial effects on the improvement in inflammation-related iron metabolism and disease severity during the first week of ICU admission but not anemia and 28-day mortality in sepsis patients with AKI.

Funding: This study was funded by the Shenzhen Key Medical Discipline Construction Fund (SZXK046) and the National Nature Science Foundation of China (81571869).

Ethics approval: The study protocol was approved by the Medical Ethics Committee of the First Affiliated Hospital of Dalian Medical University (PJ-KS-KY-2020-138), Dalian, China. A written informed consent was obtained from each participant or legal guardian at the time of initial admission.

Conflicts of interest: The authors declare no conflicts of interest.

Contributors: MMA and CXL contributed equally to this work as co-first authors. GP: concept and design; MMA: statistical analysis; PG: funding; MMA and CXL: drafting of the article; MMA and CXL: data collection. All authors revised the article and approved the final version of the article.

The supplementary files in this paper are available at http://wjem.com.cn.