Relationship between multiple inflammatory index level trajectories and 28-day mortality in patients on ECMO

doi:10.21203/rs.3.rs-5305017/v1

Background: Extracorporeal membrane oxygenation (ECMO) is a simplified cardiopulmonary bypass device that provides temporary respiratory and circulatory support and adequate recovery time for the heart and lung, but the mortality rate of acute and critically ill patients undergoing ECMO is still high. Progression of systemic inflammatory response is associated with mortality in ECMO patients. The objective of this study was to investigate the dynamic changes of various inflammatory markers and their relationship with 28-day mortality in patients with ECMO.

Methods: A retrospective cohort analysis was conducted on 200 patients receiving ECMO treatment evaluating inflammatory markers including neutrophil-to-lymphocyte ratio (NLR), systemic inflammatory response index (SIRI), procalcitonin (PCT), interleukin-6 (IL-6), and C-reactive protein (CRP) at various time points. A dynamic trajectory model was constructed, and survival differences between groups were assessed using Kaplan–Meier plots and log-rank tests. Multiple Cox proportional hazard models were built to analyze the relationship between dynamic trajectories and clinical outcomes. Causal mediation analysis was applied to determine whether changes in inflammatory trajectories mediated survival outcomes in patients on ECMO through other variables.

Results: Age, Acute Physiology and Chronic Health Evaluation (APACHE) II score, and elevated aspartate aminotransferase (AST) levels significantly impacted the 28-day survival rate (p<0.05), with higher mortality observed in patients exhibiting poor inflammatory trajectories. Kaplan–Meier survival analysis revealed that patients in the ascending (AS) group had a significantly higher risk of death than those in the stable (ST) and descending (DS) groups (log-rank p<0.001). Furthermore, multivariate Cox regression analysis identified IL-6 as the most strongly correlated inflammatory marker with mortality risk [Hazard ratio (HR) = 1.97, 95% confidence interval (CI) 1.35-2.87, p<0.001].

Conclusions: This study highlights the importance of dynamic monitoring of inflammatory biomarkers in patients on ECMO, suggesting that individualized treatment adjustments based on these markers could enhance survival rates. Future research should prioritize larger, multicenter cohort studies and clinical trials to validate these findings, aiming to optimize treatment strategies for patients on ECMO.

ECMO

inflammatory index level

trajectories

28-day mortality

interleukin-6

Extracorporeal Membrane Oxygenation (ECMO)¹ was first utilized for respiratory support outside the operating room in 1971. ECMO provides continuous extracorporeal life support, temporarily taking over the function of the heart and lungs either partially or entirely, allowing these organs to rest and recover while managing heart and lung lesions². Despite its benefits, the mortality rate for acute and critically ill patients receiving ECMO remains high³. Several factors are believed to exacerbate the inflammatory response, pushing patients into a critical state, including surgical trauma, ischemia/reperfusion injury, and endotoxin release, all of which can trigger severe inflammation⁴. The body’s inflammatory response is activated through various triggers, and if not adequately regulated, it can escalate into an overwhelming inflammatory state. This process is driven by the release of endogenous inflammatory mediators, which can intensify localized inflammation, eventually progressing to a systemic inflammatory response. When this inflammation spirals out of control, it results in systemic inflammatory response syndrome (SIRS) characterized by microcirculatory disturbances, infection, organ dysfunction, and potentially death^5,6.

Blood exposure to the extracorporeal circuit during ECMO therapy may initiate an inflammatory response akin to SIRS. However, this relationship remains poorly understood and inadequately studied⁷. Interleukin-6 (IL-6) is a critical inflammatory mediator involved in anti-inflammatory and pro-inflammatory processes through classical and trans-signaling pathways⁸. Interleukin-6 has been identified as a potential prognostic marker for mortality during ECMO therapy, as well as for complications such as lung dysfunction following cardiopulmonary bypass and acute kidney injury after cardiac surgery⁹. The Acute Physiology and Chronic Health Evaluation (APACHE)¹⁰ score is the most widely used metric for assessing disease severity in adult patients admitted to the intensive care unit (ICU). Several studies have demonstrated that the APACHE II score is a reliable predictor of mortality in patients with acute respiratory distress syndrome (ARDS) receiving ECMO¹¹. Recent research has further emphasized the role of inflammatory markers, such as the neutrophil-to-lymphocyte ratio (NLR) and the systemic inflammatory response index (SIRI), in predicting outcomes for patients with ARDS on ECMO support^12,13. Monitoring the trajectory of these inflammatory markers during ECMO treatment can offer valuable insights into patient prognosis and help inform treatment decisions. Ongoing research into the dynamic nature of these biomarkers is crucial for developing targeted interventions to mitigate the severity of this life-threatening condition.

This study aimed to explore the dynamic changes in inflammatory biomarkers in patients on ECMO and their association with 28-day mortality. It also sought to provide valuable insights that could guide clinical decision-making for patients receiving ECMO treatment by assessing the potential mediating effects of these biomarkers on survival outcomes.

2.1 Study population

This retrospective cohort study included patients on ECMO admitted to the electronic ICU (EICU) and ICU of Binhai County People's Hospital and the Second People's Hospital Affiliated with Nanjing Medical University from 2022 to 2024. The study was approved by the Ethics Committee of Binhai County People's Hospital (Approval No. 2024-BHKYLL-018). The study adhered to the ethical standards outlined in the Declaration of Helsinki.

Inclusion criteria were: 1. patients who underwent ECMO in the ICU for various conditions; 2. age ≥ 18 years; 3. ECMO duration exceeding 48 hours; and 4. Data availability regarding 28-day clinical outcomes.

Exclusion criteria included: 1. patients ≤ 18 years of age; 2. hospital stays of < 48 hours; and 3. incomplete detection of inflammatory markers at the initiation of ECMO, 24 hours into ECMO, and post-ECMO, which was required for trajectory model construction.

2.2 Clinical data collection and follow-up

Data for the study were retrieved from the hospital’s electronic medical records. The variables included age, gender, neutrophil count, lymphocyte count, monocyte count, NLR, SIRI, procalcitonin (PCT), IL-6, C-reactive protein (CRP), aspartate aminotransferase (AST), alanine aminotransferase (ALT), creatinine, urea nitrogen, APACHE-II score, underlying diseases (hypertension and diabetes), and length of hospital stay. Laboratory markers were collected by trained personnel and analyzed using a standardized, automated system under the supervision of laboratory physicians. Follow-up was conducted through outpatient visits or telephone interviews, focusing on the 28-day mortality rate among those on ECMO.

The inflammatory indices are calculated using the following formula: NLR = neutrophils/lymphocytes; SIRI = neutrophils * monocytes/lymphocytes.

2.3 Dynamic trajectory model construction for inflammatory indices

Dynamic trajectory models for the inflammatory markers were constructed using the STATA 17.0 traj package proposed by Leffondre et al.¹⁴. This analysis identified the trends of inflammatory indices across three measured time points. The best trajectory pattern was determined using relevant criteria. Locus analysis using the traj package divided each inflammatory marker into one to three subgroups, with cluster analysis grouping patients exhibiting similar dynamic changes. The primary reference parameters included^15,16: 1. Bayesian Information Criteria (BIC), a measure of model complexity and goodness-of-fit, where lower values indicate better fit; 2. Ensuring that each trajectory group comprises at least 5% of participants to maintain statistical significance; 3. The Akaike Information Criterion (AIC), akin to BIC, evaluates model fit but emphasizes fit over complexity, making it better suited for smaller sample sizes; 4. Log-Likelihood (LL), assessing the model’s fit to the observed data. 5. Entropy measures the degree of separation between trajectory groups within the model. Higher entropy values indicate more apparent group distinctions, typically from 0 to 1. A value above 0.8 is considered optimal. The Odds of Correct Classification (OCC) is used to assess the accuracy of model grouping, typically selecting five or more models. Based on these parameters, the identified trajectories were considered significant, representing the underlying patterns of change in inflammatory markers in the study population.

2.4 Causal mediation analysis of inflammatory trajectories and 28-day survival

Mediation analysis was conducted to explore whether the dynamic trajectories of inflammatory markers mediated the 28-day survival outcomes of patients on ECMO. The dose-response relationship between the inflammatory trajectory (X) and patient survival (Y) was hypothesized to be mediated by an inflammatory marker (M). The total effect (TE) of the inflammatory trajectory on survival was decomposed into direct (DE) and indirect effects (IE). According to Baron¹⁷ and Kenny¹⁸ et al., an intermediate effect of "M" is considered to be confirmed when both the association between "X" and "M" and that between "M" and "Y" are substantial.

2.5 Statistical Analysis

For continuous variables following a normal distribution, data were presented as mean ± standard deviation, with significant differences between groups assessed via t-tests. Non-normally distributed variables were expressed as interquartile ranges (IQR) and analyzed using the Wilcoxon rank-sum test. Categorical variables were presented as percentages, with comparisons conducted using chi-square tests. Patients were grouped based on the dynamic trajectory model, and survival differences were assessed using Kaplan–Meier plots and log-rank tests. Multiple Cox proportional risk models were constructed to evaluate the relationship between dynamic trajectories and clinical outcomes. Mediation analysis investigated whether dynamic inflammatory trajectories influenced survival in ECMO-supported patients via other variables. Subgroup analyses were conducted to explore potential modifiers of the relationship between dynamic trajectories and outcomes, including gender (male vs. female), age (< 60 years vs. ≥60 years), and the presence of comorbidities (yes vs. no). Statistical analyses were performed using R software (version 4.3.0) and STATA 17.0 (64-bit), utilizing R packages such as "intermediary" and "traj." Visualizations were created using "ggplot2," and bilateral p-values < 0.05 were considered statistically significant.

3.1 Baseline characteristics of participants

The study included 200 eligible patients on ECMO, of whom 83 (41.5%) survived, and 117 (58.5%) died by the 28-day follow-up. Table 1 compares the baseline characteristics of survivors and non-survivors. Older age, higher APACHE II scores, and elevated AST levels were significantly associated with increased 28-day mortality (all p < 0.05). Analysis of inflammatory markers showed no significant differences in NLR, SIRI, CRP, IL-6, and PCT at the start of ECMO treatment. However, 24 hours post-treatment and at the end of treatment, these inflammatory markers were significantly higher in the non-survivor group (all p < 0.05). No considerable differences were found in gender, creatinine, urea nitrogen, or ALT levels between the groups (all p > 0.05).

Table 1

The basic clinical characteristics of the patients were included.
Characteristics	Total No. (%)	28-day survival	28-day death	p-value
Characteristics	Total No. (%)	No. (%)	No. (%)	p-value
Total	200	83 (41.5)	117 (58.5)
Age (years)	58.00(48.00,68.00)	54.00(42.00, 64.50)	62.00(53.00, 69.00)	< 0.001
Gender, n(%)				0.758
Female	77 (38.5%)	33 (39.8%)	44 (37.6%)
Male	123 (61.5%)	50 (60.2%)	73 (62.4%)
NLR
NLR- ECMO Start	5.94(4.47,8.02)	5.78(4.74, 7.77)	6.07(4.27, 8.07)	0.861
NLR- 24 h Mark	6.30(4.23,10.12)	4.28(3.38, 5.95)	9.51(6.29, 12.36)	< 0.001
NLR- ECMO End	7.48(3.52,16.59)	3.21(2.30, 4.20)	15.56(9.75, 25.20)	< 0.001
SIRI
SIRI- ECMO Start	5.40(3.34,7.78)	5.87(3.49, 7.29)	5.32(3.34, 8.38)	0.946
SIRI- 24 h Mark	6.58(3.15,12.67)	3.13(1.81, 5.05)	11.58(6.70, 15.34)	< 0.001
SIRI- ECMO End	11.06(2.23,25.72)	1.87(1.03, 2.84)	23.71(14.13, 38.66)	< 0.001
CRP (mg/L)
CRP- ECMO Start	56.80(22.75,103.47)	57.29(27.25, 91.08)	56.06(21.00, 106.23)	0.963
CRP- 24 h Mark	73.50(36.60,118.24)	42.10(25.60, 78.00)	106.20(58.35, 140.13)	< 0.001
CRP- ECMO End	84.60(34.34,167.05)	33.35(12.25, 54.48)	138.20(87.56, 221.60)	< 0.001
IL-6 (pg/mL)
IL-6- ECMO Start	248.52(174.54,347.49)	225.60(156.78, 331.95)	265.40(186.76, 356.20)	0.050
IL-6- 24 h Mark	317.63(185.60,437.28)	162.30(103.00, 250.00)	412.32(318.90, 489.60)	< 0.001
IL-6- ECMO End	387.40(100.89,590.00)	86.90(57.66, 118.32)	566.54(426.30, 682.30)	< 0.001
PCT (ng/mL)
PCT- ECMO Start	6.54(3.18,10.28)	5.72(3.22, 10.47)	6.58(3.17, 10.25)	0.890
PCT- 24 h Mark	6.82(2.75,13.63)	3.07(1.84, 6.14)	10.65(5.88, 17.80)	< 0.001
PCT- ECMO End	4.67(1.87,15.64)	1.74(1.22, 2.43)	14.30(7.65, 20.76)	< 0.001
Creatinine (mg/dL)	99.00(78.50,145.85)	96.40(75.38, 134.00)	106.30(79.00, 151.00)	0.160
BUN (mg/dL)	8.20(6.35,10.95)	7.68(5.90, 10.18)	8.20(6.60, 11.00)	0.105
ALT (U/L)	35.60(21.60,118.80)	33.50(20.72, 95.50)	37.60(22.60, 120.60)	0.202
AST (U/L)	56.00(38.90,148.40)	48.30(38.75, 90.70)	65.60(39.80, 208.00)	0.026
APACHE II	28.00(27.00,30.00)	27.00(25.00, 29.00)	29.00(28.00, 31.00)	< 0.001
Hypertension, n(%)				0.004
No	96 (48.0%)	50 (60.2%)	46 (39.3%)
Yes	104 (52.0%)	33 (39.8%)	71 (60.7%)
Diabetes, n(%)				< 0.001
No	124 (62.0%)	63 (75.9%)	61 (52.1%)
Yes	76 (38.0%)	20 (24.1%)	56 (47.9%)
Abbreviations: NLR: Neutrophil-to-Lymphocyte Ratio; SIRI: Systemic Inflammation Response Index;
BUN: Systemic Inflammation Response Index; ALT: Alanine Aminotransferase; AST: Aspartate Aminotransferase; APACHE II: Acute Physiology and Chronic Health Evaluation II;

3.2 Construction of dynamic trajectory model of inflammation indicators

The study group performed trajectory analysis on each inflammatory indicator using the traj package, adhering to parameter criteria including BIC, AIC, LL, OCC, Entropy, and the proportion of patients in each subgroup (Supplementary Table 1). Patients were categorized into three groups: (i) ascending (AS), indicating a significant increase in inflammatory markers post-ECMO treatment; (ii) stable (ST), where the inflammatory index remained relatively constant; (iii) descending (DS), characterized by a decrease in inflammatory markers after ECMO treatment (Fig. 1A-E).

A distribution map was constructed to further investigate the distribution of patients in these groups and their baseline characteristics, depicting 28-day survival rates among deceased and surviving patients (Supplementary Fig. 1). Our analysis revealed that the ST group constituted the majority of patients, while the AS group had a higher representation than the DS group. Moreover, the study of each inflammatory indicator group (Supplementary Table 2) indicated substantial differences in 28-day survival among the three groups for NLR, SIRI, CRP, IL-6, and PCT, with the mortality rate in the AS group significantly exceeding that of the survival group (all p < 0.05).

3.3 Relationship between dynamic trajectories of inflammatory indicators and 28-day mortality in patients on ECMO

Based on the trajectory model grouping, Kaplan–Meier survival curve analysis was utilized to compare the incidence of primary outcomes across each group (Fig. 2). NLR analysis revealed that patients on ECMO in the AS group faced a significantly higher risk of 28-day mortality than those in the ST and DS groups, with the ST group demonstrating the lowest mortality risk (log-rank p < 0.001), as illustrated in Fig. 2A. Similarly, the SIRI analysis corroborated these findings, indicating that patients in the AS group experienced a significantly elevated risk of 28-day mortality compared to the ST and DS groups, with the ST group again exhibiting the lowest risk (log-rank p < 0.001), as shown in Fig. 2B. CRP analysis indicated that patients in the AS group had a significantly higher risk of 28-day mortality than those in the ST and DS groups, with the ST group having the lowest risk (log-rank p < 0.001), as presented in Fig. 2C. Subsequent IL-6 analysis revealed that the mortality risk for patients on ECMO in the DS group was intermediate between those in the ST and AS groups, with the ST group at the lowest and the AS group at the highest risk (log-rank p < 0.001), as depicted in Fig. 2D. Finally, PCT analysis confirmed that patients in the AS group had the highest mortality risk. In contrast, the ST group had the lowest, with significant differences noted compared to the other two groups (log-rank p < 0.001), as illustrated in Fig. 2E.

Moreover, the research team constructed four multifactor Cox regression models for each inflammatory indicator (Table 2) to validate the findings. In model 1, unadjusted for any variables, the analysis revealed that the risk of 28-day mortality was significantly increased in AS patients relative to the ST group (all p < 0.05). Among the indicators, IL-6 was identified as the most significant [hazard ratio (HR) = 1.97, 95% confidence interval (CI) 1.35–2.87, p < 0.001], while CRP exhibited the weakest association (HR = 1.44, 95% CI 1.12–2.26, p = 0.031). Conversely, patients in the DS group demonstrated a significant reduction in 28-day mortality risk, with SIRI presenting the weakest association (HR = 0.24, 95% CI 0.09–0.60, p = 0.002). When adjusting for age and gender in Model 2, the correlations between NLR and SIRI weakened compared to Model 1. The associations with CRP, IL-6, and PCT strengthened, and all results remained significant (p < 0.05). In Model 3, after further adjusting for comorbidities of hypertension and diabetes, the significance of SIRI disappeared, whereas the other inflammatory indicators retained their substantial associations (p < 0.05). Finally, Model 4 incorporated additional variables such as APACHE II, AST, ALT, blood urea nitrogen, and serum creatinine, again showing no significant correlation for SIRI. However, the other markers remained strongly associated with 28-day mortality. The AS group exhibited a significantly higher mortality risk compared to the ST group, with IL-6 showing the significant associations (HR = 2.48, 95% CI 1.63–3.76, p < 0.001) and NLR the weakest (HR = 1.45, 95% CI 1.01–2.23, p = 0.042). The DS group significantly reduced mortality risk, with CRP demonstrating the weakest association (HR = 0.28, 95% CI 0.14–0.57, p = 0.002). These findings are consistent with the Kaplan–Meier survival analysis trends.

Table 2

Relationship between different inflammatory indices and 28-day survival in ECMO patients.
Variable	Model 1	Model 2	Model 3	Model 4
Variable	HR (95% CI), p-value	HR (95% CI), p-value	HR (95% CI), p-value	HR (95% CI), p-value
NLR
stable	Reference	Reference	Reference	Reference
ascending	1.78 (1.20–2.66) 0.004	1.67 (1.21–2.50) 0.012	1.59 (1.06–2.39) 0.026	1.45 (1.01–2.23) 0.042
descending	0.32 (0.15–0.66) 0.002	0.33 (0.16–0.69) 0.003	0.35 (0.17–0.74) 0.006	0.35 (0.16–0.74) 0.006
SIRI
stable	Reference	Reference	Reference	Reference
ascending	1.66 (1.08–2.54) 0.021	1.60 (1.04–2.46) 0.034	1.48 (0.96–2.29) 0.076	1.30 (0.83–2.03) 0.25
descending	0.24 (0.09–0.60) 0.002	0.27 (0.11–0.66) 0.004	0.26 (0.11–0.65) 0.004	0.26 (0.10–0.64) 0.004
CRP
stable	Reference	Reference	Reference	Reference
ascending	1.44 (1.12–2.26) 0.031	1.52 (1.14–2.34) 0.029	1.50 (0.97–2.32) 0.067	1.55 (1.04–2.42) 0.048
descending	0.28 (0.14–0.54) <0.001	0.26 (0.14–0.53) <0.001	0.25 (0.13–0.49) <0.001	0.28 (0.14–0.57) <0.001
IL-6
stable	Reference	Reference	Reference	Reference
ascending	1.97 (1.35–2.87) <0.001	2.23 (1.52–3.27) <0.001	2.12 (1.45–3.12) <0.001	2.48 (1.63–3.76) <0.001
descending	0.26 (0.11–0.59) 0.001	0.27 (0.12–0.63) 0.002	0.26 (0.11–0.61) 0.002	0.29 (0.12–0.67) 0.004
PCT
stable	Reference	Reference	Reference	Reference
ascending	1.91 (1.31–2.78) 0.001	1.94 (1.33–2.84) 0.001	1.96 (1.33–2.89) 0.001	2.31 (1.54–3.46) <0.001
descending	0.26 (0.12–0.56) 0.001	0.27 (0.12–0.60) 0.001	0.28 (0.13–0.63) 0.002	0.31 (0.14–0.68) 0.004
Model 1 was a non-adjusted model.
Model 2 was adjusted for age (years), gender.
Model 3 was adjusted for the same parameters as Model 2 with additional adjustments for hypertension (Yes or No), and diabetes (Yes or No) .
Model 4 was adjusted for the same parameters as Model 3 with additional adjustments for APACHE II, AST, ALT, urea nitrogen, and serum creatinine.
HR, hazard ratio; 95% CI, 95% confidence interval. The bold values indicate statistically significant values (p < 0.05).

3.4 Dynamic trajectory of inflammatory markers and mediating causality with 28-day mortality in patients on ECMO

The progression of inflammation plays a critical role in the survival outcomes of patients on ECMO. A subsequent mediating causal analysis was conducted to understand the influence of inflammation indicator trajectories on mortality and to determine whether other factors contribute (Fig. 3). The results indicate that the dynamic trajectory of inflammatory markers predominantly mediates the 28-day mortality risk in patients on ECMO through their inflammatory responses. However, other factors also contribute to mortality outcomes, highlighting the complexity of this relationship. Analysis of NLR dynamic trajectories revealed that NLR-mediated effects accounted for 57.5% of the association between NLR trajectories and 28-day mortality risk (IE = 0.387, 95% CI: 0.552–0.792; DE = 0.28, 95% CI: 0.509–0.82; Fig. 3A). SIRI-mediated effects contributed to 78.64% of the association between SIRI trajectories and 28-day mortality (IE = 0.663, 95% CI: 0.552–0.792; DE = 0.005, 95% CI: = 0.276–1.030; Fig. 3B). CRP-mediated effects accounted for 62.7% of the association between CRP trajectories and mortality risk (IE = 0.859, 95% CI: 0.055–0.680; DE = 0.329, 95% CI: = 0.120–0.530; Fig. 3C). Interleukin-6-mediated effects explained 61.2% of the association between IL-6 trajectories and 28-day mortality risk (IE = 0.745, 95% CI: 0.570–0.924; DE = 0.132, 95% CI: = 0.014–1.280; Fig. 3D). PCT-mediated effects contributed 42.1% of the association between PCT trajectories and mortality (IE = 0.270, 95% CI: 0.170–0.586; DE = 0.362, 95% CI: 0.170–0.533; Fig. 3E). These findings indicate that, while SIRI exhibited the highest mediation effect and PCT the lowest, other underlying mechanisms beyond inflammatory indicators likely influence patient mortality, as the relationship is not solely dependent on these markers.

3.5 Subgroup analysis

Subgroup analyses were conducted based on demographic and clinical parameters, including age, gender, and comorbidities (diabetes and hypertension) (Fig. 4). Among men, the AS group exhibited a significantly higher 28-day mortality risk compared to other groups, while no significant differences were observed among women. However, the DS group had the lowest mortality risk for women. Patients < 60 years of age showed the highest mortality risk in the AS group and the lowest risk in the DS group, consistent with the primary analysis. Similarly, patients without hypertension or diabetes also displayed the highest mortality risk in the AS group and the lowest in the DS group. Further analysis of SIRI, CRP, IL-6, and PCT (Fig. 4B-E) revealed varying risks associated with different inflammatory marker trajectories across patient subpopulations. These results suggest that specific patient characteristics influence the dynamic trajectories of inflammatory markers, aligning with the results from the mediating causal analysis and indicating that additional factors likely intervene in the survival outcomes of patients on ECMO.

ECMO primarily serves to provide respiratory and circulatory support to critically ill patients as a bridge to recovery, particularly for those who otherwise face fatal outcomes¹⁹. As an extension of cardiopulmonary bypass, ECMO is not without complications, such as inflammation, microcirculatory disturbances, and blood cell damage²⁰. Among older patients receiving ECMO, particularly those with elevated APACHE II scores and AST levels, the 28-day mortality risk was significantly higher. Although the APACHE II scoring system is commonly used to predict outcomes in patients on ECMO, it tends to underestimate mortality in low-risk patients and overestimate it in high-risk patients²¹. Recent studies have validated that elevated liver enzymes, such as AST/ALT levels > 70 units/L, are indicative of liver injury and correlate with increased mortality in patients with ECMO²².

Although our initial analysis of inflammatory markers (NLR, SIRI, CRP, IL-6, and PCT) showed no significant differences at the beginning of ECMO treatment, substantial differences emerged after 24 hours and at treatment conclusion. Notably, these markers were significantly higher in non-survivors compared to survivors. Our study provides a novel contribution to the existing literature by exploring the dynamic changes in these inflammatory markers among patients undergoing ECMO therapy. Biomarkers of inflammation, endothelial activation, and fibrinolytic activity undergo significant shifts within minutes to hours following ECMO initiation²³. This investigation addresses a critical knowledge gap by highlighting the prognostic value of these biomarkers for predicting outcomes in critically ill patients, reinforcing their potential as predictive tools in clinical practice.

Activating ECMO exacerbates inflammation and coagulation abnormalities, which are closely intertwined and complex²³. Inflammatory responses play a central role in the progression of clinical outcomes in patients receiving ECMO. Kaplan–Meier survival curve analysis of NLR, SIRI, CRP, IL-6, and PCT revealed that the AS group faced the highest 28-day mortality risk, while those in the ST group had the lowest. The multivariate Cox regression model confirmed a significantly elevated 28-day mortality risk in the AS group, with IL-6, NLR, CRP, IL-6, and PCT being key contributors, among which IL-6 was the most prominent marker. Elevated plasma concentrations of pro-inflammatory cytokines, including tumor necrotic factor-alpha (TNF-α), IL-8, IL-1β, and IL-6, have been observed early after ECMO initiation²⁴. IL-6 has been identified as a valuable prognostic marker for mortality during ECMO support, lung dysfunction after cardiopulmonary bypass, and acute kidney injury following cardiac surgery⁶. It triggers the production of acute-phase reactants, including CRP and fibrinogen, while reducing albumin synthesis²⁵. Recent studies emphasize the predictive significance of inflammatory measures such as NLR and SIRI in patients with ARDS receiving ECMO support^12,13. Our analysis suggests that the dynamic trajectories of these inflammatory markers largely influence the 28-day mortality risk in patients on ECMO, though additional factors may also play a role. In this association, the mediation efficacy of NLR, SIRI, CRP, IL-6, and PCT were 57.5%, 78.64%, 62.7%, 61.2%, and 42.1%, respectively.

Furthermore, the dynamic changes in inflammatory indicators varied based on patient characteristics. For instance, NLR analysis showed that in men < 60 years of age without hypertension or diabetes, the 28-day mortality risk was significantly higher in the AS group compared to others. SIRI analysis indicated that men < 60 years of age in the AS group faced a significantly higher 28-day mortality risk compared to those in the remaining groups. Similarly, CRP analysis revealed that women with hypertension in the AS group exhibited a markedly increased risk of death within 28 days compared to their counterparts. Moreover, assessments of IL-6 and PCT showed that patients in the AS group had a significantly higher 28-day mortality risk compared to those in other groups.

These findings hold important implications for clinical practice and policy-making. The correlation between elevated inflammatory markers and increased mortality risk suggests that continuous monitoring of these indices could guide clinicians in making timely and informed patient management decisions. Early identification of patients at heightened risk could facilitate more personalized treatment strategies, such as intensified anti-inflammatory interventions or tailored ECMO settings, potentially improving survival outcomes. Furthermore, this research emphasizes the necessity of incorporating routine assessments of inflammatory biomarkers into existing ECMO management protocols, which could improve patient care and optimize resource allocation within healthcare systems.

However, several limitations of this study must be acknowledged. The relatively small sample size may impede the generalizability of our findings, as a larger cohort could yield more robust evidence regarding the association between inflammatory markers and 28-day mortality in patients on ECMO. Moreover, the retrospective nature of the analysis may introduce biases that affect the observed associations. Future research should strive to replicate these findings in more extensive, multicenter cohorts, evaluating the predictive value of inflammatory indicators across diverse clinical settings. Furthermore, the absence of long-term follow-up data restricts our understanding of the chronic implications of inflammatory marker dynamics in patients post-ECMO. Addressing these limitations is essential for advancing our knowledge and enhancing outcomes in this vulnerable population.

This research underscores the significant relationship between the dynamic changes in inflammatory markers and 28-day mortality in ECMO patients, highlighting the critical role of biomarkers in prognostic assessment. The findings advocate the integration of continuous monitoring of validated trajectories into clinical practice, which may help to adjust treatment in a timely manner and improve patient outcomes. By identifying specific inflammatory trajectories that correlate with worse prognosis, this study lays the groundwork for the development of personalized treatment strategies aimed at enhancing survival rates in this vulnerable patient population. Future investigations should focus on refining these prognostic models and exploring the potential for targeted interventions based on inflammatory responses.

Ethics approval and consent to participate

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of Binhai County People's Hospital (Approval No. 2024-BHKYLL-018).

Consent for publication

All the authors of the article agree to participate in the journal submission. Data was released with patient consent.

Availability of data and material

The datasets generated and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Competing interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Authors' Contributions

YZ and ZC substantially contributed to the conception and design of the study. LQS and JYH performed the research and prepared the figures for the study. YZ and SHL analyzed and interpreted the data and drafted the manuscript. YZ, ZC, and LQS critically revised the important intellectual content of the manuscript and provided final approval. LMX and XW supervised the study. All authors read and approved the final manuscript.

Acknowledgments

Thanks to all the collaborators of this paper for their data analysis and paper writing.

Luciano G, Eleonora C, Thomas L. Clinical review: Extracorporeal membrane oxygenation. Crit Care. 2011;15(6):0.
Schmidt M, Hodgson C, Combes A. Extracorporeal gas exchange for acute respiratory failure in adult patients: a systematic review. Crit Care. Mar 16 2015;19(1):99.
Aso S, Matsui H, Fushimi K, Yasunaga H. In-hospital mortality and successful weaning from venoarterial extracorporeal membrane oxygenation: analysis of 5,263 patients using a national inpatient database in Japan. Crit Care. Apr 5 2016;20:80.
D P, T M Y, E Y. Cardiopulmonary bypass induced inflammation: pathophysiology and treatment. An update. . Eur J Cardiothorac Surg. 2002;21(2):0.
Millar JE, Fanning JP, McDonald CI, McAuley DF, Fraser JF. The inflammatory response to extracorporeal membrane oxygenation (ECMO): a review of the pathophysiology. Crit Care. Nov 28 2016;20(1):387.
Angus DC, van der Poll T. Severe sepsis and septic shock. N Engl J Med. Aug 29 2013;369(9):840-851.
Bernardi MH, Rinoesl H, Dragosits K, et al. Effect of hemoadsorption during cardiopulmonary bypass surgery - a blinded, randomized, controlled pilot study using a novel adsorbent. Crit Care. Apr 9 2016;20:96.
Hasegawa H, Mizoguchi I, Chiba Y, Ohashi M, Xu M, Yoshimoto T. Expanding Diversity in Molecular Structures and Functions of the IL-6/IL-12 Heterodimeric Cytokine Family. Front Immunol. 2016;7:479.
P D, C A, J K, et al. Urine interleukin-6 is an early biomarker of acute kidney injury in children undergoing cardiac surgery. Crit Care. 2010;14(5).
Salluh JI, Soares M. ICU severity of illness scores: APACHE, SAPS and MPM. Curr Opin Crit Care. Oct 2014;20(5):557-565.
Auriemma CL, Zhuo H, Delucchi K, et al. Acute respiratory distress syndrome-attributable mortality in critically ill patients with sepsis. Intensive Care Med. Jun 2020;46(6):1222-1231.
Yoshino A, Nakamura Y, Hashiguchi S, et al. The Association between the Oral-Gut Axis and the Outcomes of Severe COVID-19 Patients Receiving Extracorporeal Membrane Oxygenation: A Case-Control Study. J Clin Med. Feb 22 2022;11(5).
Thangappan K, Cavarocchi NC, Baram M, Thoma B, Hirose H. Systemic inflammatory response syndrome (SIRS) after extracorporeal membrane oxygenation (ECMO): Incidence, risks and survivals. Heart Lung. Sep-Oct 2016;45(5):449-453.
Leffondre K, Abrahamowicz M, Regeasse A, et al. Statistical measures were proposed for identifying longitudinal patterns of change in quantitative health indicators. J Clin Epidemiol. Oct 2004;57(10):1049-1062.
Nguena Nguefack HL, Page MG, Katz J, et al. Trajectory Modelling Techniques Useful to Epidemiological Research: A Comparative Narrative Review of Approaches. Clin Epidemiol. 2020;12:1205-1222.
Daniel S N, Bobby L J, Jonathan E. Recent Advances in Group-Based Trajectory Modeling for Clinical Research. Annu Rev Clin Psychol. 2024;20(1):0.
R M B, D A K. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol. 1986;51(6):1173.
He A, Liu J, Qiu J, et al. Risk and mediation analyses of hemoglobin glycation index and survival prognosis in patients with sepsis. Clin Exp Med. Aug 7 2024;24(1):183.
Abrams D, MacLaren G, Lorusso R, et al. Extracorporeal cardiopulmonary resuscitation in adults: evidence and implications. Intensive Care Med. Jan 2022;48(1):1-15.
Liu M, Li X, Zhou R. Severe coagulopathy and inflammation occurred after resection of giant right ventricular intimal sarcoma with cardiopulmonary bypass: a case report. BMC Anesthesiol. Jan 31 2024;24(1):43.
Ng WT, Ling L, Joynt GM, Chan KM. An audit of mortality by using ECMO specific scores and APACHE II scoring system in patients receiving extracorporeal membrane oxygenation in a tertiary intensive care unit in Hong Kong. J Thorac Dis. Feb 2019;11(2):445-455.
Masha L, Peerbhai S, Boone D, et al. Yellow Means Caution: Correlations Between Liver Injury and Mortality with the Use of VA-ECMO. ASAIO J. Nov/Dec 2019;65(8):812-818.
Caprarola SD, Ng DK, Carroll MK, et al. Pediatric ECMO: unfavorable outcomes are associated with inflammation and endothelial activation. Pediatr Res. Aug 2022;92(2):549-556.
Panigada M, Spinelli E, De Falco S, et al. The relationship between antithrombin administration and inflammation during veno-venous ECMO. Sci Rep. Aug 22 2022;12(1):14284.
Tanaka T, Narazaki M, Kishimoto T. IL-6 in inflammation, immunity, and disease. Cold Spring Harb Perspect Biol. Sep 4 2014;6(10):a016295.

No competing interests reported.

STablesFigures.docx

Relationship between multiple inflammatory index level trajectories and 28-day mortality in patients on ECMO

Status:

Version 1

Abstract

Figures

1. Introduction

2. Methods

2.1 Study population

2.2 Clinical data collection and follow-up

2.3 Dynamic trajectory model construction for inflammatory indices

2.4 Causal mediation analysis of inflammatory trajectories and 28-day survival

2.5 Statistical Analysis

3. Results

3.1 Baseline characteristics of participants

3.2 Construction of dynamic trajectory model of inflammation indicators

3.3 Relationship between dynamic trajectories of inflammatory indicators and 28-day mortality in patients on ECMO

3.4 Dynamic trajectory of inflammatory markers and mediating causality with 28-day mortality in patients on ECMO

3.5 Subgroup analysis

4. Discussion

5. Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1