Predictive value of lymphocyte‐to‐monocyte ratio in critically Ill patients with atrial fibrillation: A propensity score matching analysis

Abstract Background Inflammation plays a key role in the initiation and progression of atrial fibrillation (AF). Lymphocyte‐to‐monocyte ratio (LMR) has been proved to be a reliable predictor of many inflammation‐associated diseases, but little data are available on the relationship between LMR and AF. We aimed to evaluate the predictive value of LMR in predicting all‐cause mortality among AF patients. Methods Data of patients diagnosed with AF were retrieved from the Medical Information Mart for Intensive Care‐III (MIMIC‐III) database. X‐tile analysis was used to calculate the optimal cutoff value for LMR. The Cox regression model was used to assess the association of LMR and 28‐day, 90‐day, and 1‐year mortality. Additionally, a propensity score matching (PSM) method was performed to minimize the impact of potential confounders. Results A total of 3567 patients hospitalized with AF were enrolled in this study. The X‐tile software indicated that the optimal cutoff value of LMR was 2.67. A total of 1127 pairs were generated, and all the covariates were well balanced after PSM. The Cox proportional‐hazards model showed that patients with the low LMR (≤2.67) had a higher 1‐year all‐cause mortality than those with the high LMR (>2.67) in the study cohort before PSM (HR = 1.640, 95% CI: 1.437–1.872, p < 0.001) and after PSM (HR = 1.279, 95% CI: 1.094–1.495, p = 0.002). The multivariable Cox regression analysis for 28‐day and 90‐day mortality yielded similar results. Conclusions The lower LMR (≤2.67) was associated with a higher risk of 28‐day, 90‐day, and 1‐year all‐cause mortality, which might serve as an independent predictor in AF patients.


| INTRODUC TI ON
Atrial fibrillation (AF) is the most common sustained and supraventricular arrhythmia, characterized by uncoordinated atrial electrical activation and consequently ineffective atrial contraction. 1 AF is associated with substantial morbidity and mortality, thus posing a significant burden to patients, physicians, and healthcare systems globally. 2 Preventing AF recurrence (via rhythm control) and detrimental complications (via rate control and antithrombotic therapies) are current therapeutic strategies for AF patients. 3 The pathophysiology of AF is complex and incompletely understood. Emerging evidence suggests that the roles of activated inflammatory cells and mediators in cardiac tissue and circulatory system have been implicated in various AF-related pathological mechanisms. 4,5 The lymphocyte-to-monocyte ratio (LMR), comprised of the ratio of white blood cell (WBC) subgroups, has been proved to be a novel inflammatory marker for lots of cardiovascular diseases, such as acute type A aortic dissection (AAAD), 6 ST-elevated myocardial infarction (STEMI), 7 heart failure, 8 acute pulmonary embolism, 9 and carotid artery stenosis. 10 Several histological studies of AF found that increased infiltration of inflammatory cells, such as lymphocytes and monocytes, in the atrial myocardium or appendage tissues. [11][12][13] Another study demonstrated that a higher percentage of activated T lymphocytes was observed in the peripheral blood of patients with paroxysmal or persistent AF. 14 Furthermore, monocyte infiltration in the left atria was reported to be associated with AF-related thromboembolic events. 15,16 Nevertheless, to the best of our knowledge, there is almost no study investigating the association between LMR in the peripheral blood and the survival of AF patients.
In the present study, we intended to investigate whether there was a relationship between LMR and prognosis in critically ill patients with AF by utilizing the Medical Information Mart for Intensive Care-III (MIMIC-III) database. This research was conducted consistent with the requirements of the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) statement. 17 2 | MATERIAL S AND ME THODS

| Study design and data resource
We conducted a longitudinal, single-center retrospective cohort study with all the relevant data collected from the MIMIC-III database based on the methods used in our previous studies. 18 -20 The

| Patient selection
We included all intensive care unit (ICU) patients (aged ≥ 18 years) in the database with the primary diagnosis of AF using the ICD-9 diagnosis code (ICD-9 code of AF = 42731). Only the data of each patient's first ICU admission were used in this study. Patients were excluded if they had (1) a secondary diagnosis of inflammatory, hematological or autoimmune diseases, sepsis, or malignant tumors; (2) incomplete follow-up information; (3) a length of stay in the ICU less than 24 hours; (4) incomplete or unobtainable data of measured lymphocyte or monocyte count during the first 24-hour admission; or (5) more than 10% of individual data missing. and sleep apnea. Vital signs on admission included heart rate, respiratory rate, systolic blood pressure (SBP), diastolic blood pressure (DBP), and mean blood pressure (MBP). Laboratory-based data included WBC, neutrophil, lymphocyte, platelet, monocyte, hematocrit, hemoglobin, red blood cell distribution width (RDW), albumin, blood urea nitrogen (BUN), creatinine, glucose, total calcium (tCa), potassium, sodium, chloride, magnesium, prothrombin time (PT), partial thromboplastin time (PTT), and international normalized ratio (INR). If participants underwent more than one laboratory test during their hospitalization, only the initial test results were included for further analysis. In terms of scoring systems, the Simplified Acute Physiology Score II (SAPS II) and the Sequential Organ Failure Assessment (SOFA) were extracted from the database. Additionally, treatment information data included mechanical ventilation, renal replacement treatment, appendage closure, coronary artery bypass grafting (CABG), valvular surgery, and in-hospital medication administration (antiarrhythmic agents, antiplatelet agents, warfarin, and beta-blocker).
Our primary study outcome was 1-year all-cause mortality. The secondary outcomes included 28-day and 90-day all-cause mortality.

| Definition, calculation, and identification of cutoff values for LMR
Lymphocyte-to-monocyte ratio was calculated in the formulate: lymphocyte counts divided by monocyte counts on admission. LMR, as a continuous variable, was dichotomized via the X-tile software (version 3.6.1; Yale University, New Haven, CT, USA) based on the maximal log-rank chi-square value, which represented the greatest group difference in outcome probability. 22 In addition, normal ranges of lymphocyte and monocyte counts in the peripheral blood were defined as between 0.8 × 10 9 /L and 4.0 × 10 9 /L, and between 0.12 × 10 9 /L and 0.8 × 10 9 /L, respectively.

| Management of missing data
To reduce bias due to missing data, variables with more than 20% missing values were excluded from the study. Correspondingly, variables with less than 20% missing values were handled using multivariable imputation. 23 Variables for which multivariable imputation was adopted included RDW, BUN, tCa, chloride, PT, PTT, and INR.

| Propensity score matching
Propensity score matching (PSM) analysis was used to minimize the effect of potential confounders. Baseline characteristics (age, sex, current smoking status, admission type, CAD, congestive heart failure, hypertension, COPD, stroke, TIA, DM, dyslipidemia, anemia, chronic kidney disease, chronic liver disease, sleep apnea, SBP, DBP, MBP, heart rate, respiratory rate, WBC, neutrophil, platelet, hematocrit, hemoglobin, RDW, albumin, BUN, creatinine, tCa, potassium, sodium, chloride, magnesium, PT, PTT, INR, SOFA, SAPS II, mechanical ventilation, renal replacement treatment, appendage closure, CABG, valvular surgery, and in-hospital medication administration) were incorporated in the propensity score analysis. We did not include lymphocyte and monocyte counts in the PSM analysis to avoid influence on the value of LMR. A logistic regression model was constructed to calculate and assign each patient a propensity score, which was defined as the likelihood of being exposed to an intervention given that the status of a particular patient's measured prognostic factors. 24 presented as hazard ratios (HRs) and 95% confidence intervals (CIs).
The LMR > 2.67 group was taken as the reference group. We also did the subgroup analysis based on lymphocyte and monocyte counts, age, sex, CAD, congestive heart failure, hypertension, COPD, stroke, TIA, DM, dyslipidemia, anemia, chronic kidney disease, chronic liver disease, sleep apnea, mechanical ventilation, CABG, renal replacement treatment, and in-hospital medication administration.
Furthermore, to identify a non-linear relationship, a smooth curve was then drawn to estimate the relationship between LMR and its HR using restricted cubic spline regression analysis. Two piecewise Cox proportional-hazards models were further performed to demonstrate the saturation effect of LMR on mortality. The inflection point was determined using the recursive method, where the model gave the maximum likelihood. Furthermore, a log-likelihood ratio test comparing the one-line linear model with two piece-wise models was conducted to determine whether the saturation effect existed.
A two-tailed p < 0.050 was considered to be statistically significant. All statistical analyses were conducted using R software

| Characteristics of patients
In total, 3567 patients fulfilled the selection criteria and comprised the final study cohort (Figure 1). X-tile software identified the optimal cutoff value of LMR for 1-year mortality as 2.67. Therefore, patients were divided into the low LMR group (n = 1766) and the high LMR group (n = 1801). The baseline characteristics of enrolled patients are briefly summarized in Table 1. Patients with the higher LMR (>2.67) tended to be younger (p < 0.001). Regarding comorbidity, patients with the higher LMR (>2.67) were more likely to suffer from CAD (p = 0.002), hypertension (p < 0.001), stroke (p = 0.028), and dyslipidemia (p < 0.001). However, patients with the lower LMR (≤2.67) displayed higher WBC (p < 0.001), neutrophil (p < 0.001), SAPS II (p < 0.001); they were also more likely to receive renal replacement treatment (p < 0.001).  Figure 2A. The curves of the LMR groups differed significantly, and patients in the low LMR group had a higher cumulative incidence of mortality (log-rank test: p < 0.001).

| Prognostic significance of LMR before PSM
The results of the univariable and multivariable Cox regression analyses are summarized in Table 2 and Tables S1-3. A univariable Cox regression analysis was conducted to select the variables with p < 0.100, and age, gender, CAD, congestive heart failure, hypertension, COPD, stroke, dyslipidemia, chronic kidney disease, chronic liver disease, sleep apnea, mechanical ventilation, renal replacement treatment, appendage closure, CABG, valvular surgery, antiarrhythmic, antiplatelet agents, warfarin, and beta-blocker were selected and incorporated into the multivariable Cox regression model. Multivariable Cox regression analysis showed that patients with the LMR ≤ 2.67 had significantly higher 1-year mortality compared to patients with the LMR > 2.67 (Model 1: HR = 1.950, 95% CI: 1.713-2.220, p < 0.001; Model 2: HR = 1.640, 95% CI: 1.437-1.872, p < 0.001). The multivariable analysis for 28-day and 90-day mortality yielded similar results.

| Prognostic significance of LMR after PSM
In total, 1127 pairs of propensity score-matched patients were generated after using a 1:1 ratio PSM analysis to balance the potential confounders. The patients' baseline characteristics after PSM are il-

| Prognostic significance of LMR in patients with normal lymphocyte and monocyte counts
Considering a reduced lymphocyte count or elevated monocyte count might cause a lower LMR, which could influence the study results independently, the correlation between LMR and mortality was also analyzed in AF patients with normal lymphocyte and monocyte counts. Kaplan-Meier curves for all-cause death according to the LMR groups are shown in Figure 2B.

| Subgroup analysis
To further validate the robustness of our findings, we performed subgroup analyses to assess the association between LMR and 28day, 90-day, and 1-year all-cause mortality. For 1-year mortality, subgroup analyses showed the lower LMR was also associated with deteriorative mortality in most strata except in patients with chronic liver disease (p = 0.065), sleep apnea (p = 0.095), or receiving renal replacement treatment (p = 0.077) or CABG (p = 0.156) ( Figure S3).
The results for 28-day and 90-day mortality were shown in Figures S1-2.

| DISCUSS ION
Our study investigated the association between admission LMR in the peripheral blood and risk of death among critically ill patients with AF with a 1-year follow-up. Our findings showed that that the lower LMR (≤2.67) was associated with a higher risk of 28-day, 90day, and 1-year all-cause mortality and might serve as a reliable predictor of mortality in AF patients. As far as we know, this is the first research to explore the correlation between LMR and mortality of AF patients.
A considerable number of clinical studies have suggested that LMR could serve as an indispensable prognostic predictor in many cardiovascular diseases such as AAAD 6 , STEMI (7), heart failure (8), acute pulmonary embolism (9), and carotid artery stenosis (10).
Moreover, one recent study suggested that a preoperative lower LMR (<3.58) was associated with a higher risk of 4-year mortality in patients undergoing cardiac surgery. 27 To date, several circulating Note: CABG, coronary artery bypass grafting; CAD, coronary artery disease; CI, confidential interval; COPD, chronic obstructive pulmonary disease; HR, hazard ratio; LMR, lymphocyte-to-monocyte ratio; PSM, propensity score matching. a Adjusted model 1 was adjusted by age and sex.
c The LMR >2.67 group was taken as the reference group.
blood cell-based prognostic biomarkers have also been developed to predict clinical outcomes in AF. An elevated neutrophilto-lymphocyte ratio (NLR) before or after catheter ablation was associated with increased AF recurrence after the procedure. [28][29][30] Gungor et al. 31 and Saskin et al. 32  The present study was the first to explore the relationship between LMR and mortality among AF patients. We found that the lower LMR (≤2.67) was associated with a higher risk of 28-day, 90day, and 1-year all-cause mortality in AF patients. A PSM analysis was performed to minimize the impact of potential confounders.
The major results before and after PSM were consistent in this study. However, the values of HRs on mortality after PSM were reduced compared with those before PSM, which might be due to not only the balance of baseline characteristics but also the vari- and AF have found that elevated inflammatory cell counts including lymphocytes and monocytes in human tissue samples. [11][12][13]39,40 One recent research found a correlation between the complement system activation and lymphocyte pro-inflammatory cytokines release with the cardiac abnormalities (conduction disturbances and atrial fibrosis/remodeling). 41 Cluster of differentiation CD4 + T lymphocytes without the surface-antigen (protein) CD28, the so-called CD4 + CD28 null T cells, are reported to be involved in chronic inflammatory processes, which might impact the development and progression of AF. 42 Additionally, lymphopenia might indicate that the immune response is suppressed and this condition has been associated with adverse cardiac outcomes. Low relative lymphocyte count has been demonstrated to be associated with poor prognosis in patients with heart failure, 43 acute coronary syndromes, 35 cardiac arrest, 44 or stable coronary heart disease. 45 Furthermore, monocytes attach to adhesion molecules, proceeding into the sub-endothelial F I G U R E 3 Restricted cubic spline fitting for the association between LMR levels with the HR of LMR for 28-day (A), 90-day (B), 1-year (C) mortality. HRs were evaluated by setting the LMR value=2.67 as reference based on multivariable Cox proportional regression model adjusted by age, gender, coronary artery disease, congestive heart failure, hypertension, COPD, stroke, dyslipidemia, chronic kidney disease, chronic liver disease, sleep apnea, mechanical ventilation, renal replacement treatment, appendage closure, CABG, valvular surgery, antiarrhythmic, antiplatelet agents, warfarin, betablocker. The shaded area represents the 95% CI. CABG, coronary artery bypass grafting; CI, confidential interval; COPD, chronic obstructive pulmonary disease; LMR, lymphocyte-to-monocyte ratio space of the valve in response to locally produced cytokines such as tumor necrosis factorα and interleukin-6, which might be attributed to the mechanism of AF occurrence. 46 Abnormal changes in systemic inflammation have been related to prothrombotic indices in AF. These mechanisms might be associated with hypercoagulation, platelet activation, and endothelial dysfunction. 5 For example, monocytes could actively bind to platelets, thus forming prothrombotic monocyte-platelet aggregates, which might be involved in the process of atrial thrombus formation and associated with a worse prognosis in ischemic events. 15,16 The LMR integrates the clinical significance of lymphocytes and monocytes, and the underlying mechanisms might be related to the impact of low lymphocyte counts and high monocyte counts on the prognosis of AF. Additional studies are needed to investigate the exact mechanism.
Atrial fibrillation is the most common arrhythmia observed in clinical practice and a significant contributor to cardiovascular morbidity and possibly mortality. 47

| CON CLUS IONS
To sum up, our study results suggested that the lower LMR (≤2.67) was correlated with a higher risk of 1-year mortality among AF. The LMR could serve as a potential prognostic predictor of all-cause mortality in AF patients.

ACK N OWLED G EM ENT
We thank all the staff working in ICU of the Beth Israel Deaconess Medical Center, Boston, USA.

PATI E NT A N D PU B LI C I N VO LV E M E NT
Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

PATI E NT CO N S E NT FO R PU B LI C ATI O N
Not required.

R E P O RTI N G CH ECK LI S T
The authors have completed the STROBE reporting checklist.