Monocyte-to-HDL ratio predicts 30-day mortality in middle-aged and elderly patients with HBV-related decompensated cirrhosis

Abstract

Objective

To evaluate the prognostic value of monocyte-to-high-density lipoprotein cholesterol ratio (MHR) for 30-day mortality in middle-aged and elderly patients with hepatitis B virus-related decompensated cirrhosis (HBV-DeCi).

Methods

This single-center and retrospective cohort study analyzed 244 middle-aged and elderly patients with HBV-DeCi. We compared routine blood parameters, coagulation, hepatic, and renal function parameters between survivors and non-survivors. Feature selection was performed using LASSO regression, with independent risk factors subsequently identified through multivariate logistic regression analysis.

Results

Among the cohort (median age 55 years), non-survivors exhibited significantly elevated MHR (monocyte-to-HDL ratio) levels (median 2.62 vs. 0.43, p < 0.001). Multivariate analysis confirmed MHR as an independent 30-day mortality predictor (Odds Ratio (OR) = 1.862, 95% confidence interval (CI): 1.354–2.561, p < 0.001). An integrated model combining the Model for End-Stage Liver Disease (MELD) score with MHR demonstrated superior predictive performance (area under the curve (AUC) = 0.917) in this middle-aged and elderly population, though the difference did not reach statistical significance (DeLong test, p > 0.05).

Conclusion

MHR serves as a potential prognostic biomarker in middle-aged and elderly HBV-DeCi patients. The Model for End-Stage Liver Disease (MELD) + MHR model provides enhanced risk stratification for this vulnerable age group, which may facilitate early intervention and optimize resource allocation.

Introduction

Hepatitis B Virus (HBV) remains the leading cause of cirrhosis and liver-related mortality in Asia-Pacific populations, even in the era of modern antiviral therapy [1]. In patients with HBV-related decompensated cirrhosis (HBV-DeCi), multiple complications collectively determine clinical outcomes [2]. Although liver transplantation is the definitive treatment, its application is limited by donor scarcity and procedural demands. Therefore, early identification of high-risk patients is crucial for optimizing clinical management.

Several readily available prognostic markers have been identified for mortality in HBV-DeCi, including the neutrophil-to-lymphocyte ratio (NLR) [3], neutrophil-to- high-density lipoprotein cholesterol (HDL) ratio (NHR) [2], aspartate aminotransferase-to-platelet ratio index (APRI) [4], and the fibrosis-4 (FIB-4) index [5]. Yet these tools present drawbacks: some exhibit inter-observer variation, and many demonstrate limited sensitivity during early decompensation. Thus, a need remains for simple, objective, and widely available biomarkers.

Emerging evidence implicates persistent systemic inflammation and immune dysregulation as central drivers of cirrhotic progression, closely linked to complications and mortality [6–7]. The monocyte-to-HDL ratio (MHR) integrates inflammatory and metabolic pathways, potentially offering a more comprehensive assessment of the “inflammation-metabolism” imbalance in cirrhosis. Although MHR has been validated in atherosclerotic diseases and more recently in HBV-DeCi populations, existing studies have not addressed whether its prognostic utility varies across age strata [8–9]. This is a critical gap, as middle-aged and elderly patients—who experience more rapid disease progression and higher short-term mortality—represent a subgroup that could derive the greatest benefit from precise risk stratification.

Therefore, this study aims to evaluate the prognostic value of MHR for 30-day mortality in middle-aged and elderly patients with HBV-DeCi, and to determine whether MHR provides incremental predictive value beyond established clinical scores such as MELD.

Materials and methods

Patients

This single-center and retrospective cohort analysis utilized clinical data from middle-aged (45–64 years) and elderly (≥ 65 years) patients with HBV-DeCi admitted to the hospital between January 2017 and June 2020. Diagnosis required three criteria: (1) Persistent positivity for hepatitis B surface antigen for > 6 months. (2) Cirrhosis confirmed by clinical, imaging, or histopathological findings, accompanied by at least one of the following: (a) Ascites: Confirmed by physical examination and imaging. (b) Variceal Hemorrhage (endoscopically proven). (c) Hepatic Encephalopathy: Diagnosed according to the West-Haven criteria [10]. 3) Acute Decompensation events: Presence of the above complications or their direct sequelae (e.g., hepatorenal syndrome, spontaneous bacterial peritonitis, or new-onset jaundice). Exclusion criteria were as follows: age under 45 years, malignancy, hematological disorders, concurrent chronic liver diseases (such as hepatitis C co-infection or autoimmune hepatitis), recent platelet transfusion or immunosuppressive therapy, major cardiovascular events within the preceding three months, chronic use of medications affecting lipid metabolism or leukocyte/monocyte counts (e.g., statins, fibrates); or substantial missing clinical data. Patients with substantial missing data (defined as > 10% missing for key variables) were excluded from the analysis. For the remaining variables with missing values, random forest imputation was as applied for variables with < 10% missingness. The study was fulfilled in line with the principles presented in the Declaration of Helsinki, and was approved by the local ethics committee of the First Affiliated Hospital with Nanjing Medical University (2023-SR-022). The institutional Ethics Committee approved the research protocol and granted a waiver for individual informed consent given the retrospective design and no identifiable data were used. (Approval number: 2023-SR-022, date:8 February, 2023).

Clinical and laboratory data were systematically abstracted from electronic health records within the initial 24 h of hospitalization. Collected parameters included demographic characteristics (age and sex), complete blood counts (leukocytes (WBC), neutrophils (NE), lymphocytes (LY), monocytes (MO), hemoglobin (HB), platelet (PLT)), comprehensive biochemical profiles (alanine aminotransferase (ALT), aspartate aminotransferase (AST), direct bilirubin (DBIL), HDL, low-density lipoprotein cholesterol (LDL), albumin (ALB), urea (UREA), creatinine (CREA), sodium (Na), Model for End-Stage Liver Disease (MELD), aspartate aminotransferase to platelet ratio index (APRI), fibrosis-4 index (FIB-4)), and coagulation (prothrombin time (PT), international normalized ratio (INR), activated partial thromboplastin time (APTT), thrombin time (TT), D-dimer (D-D), and fibrinogen (FIB)). MHR (continuous) was derived from absolute monocyte count and HDL concentration. For comparative prognostic assessment, established scoring systems (MELD, APRI, and FIB-4) were computed using conventional formulas. Patient outcomes were stratified by 30-day all-cause mortality into survivor and non-survivor groups.

Statistical computations employed SPSS (26.0) with non-normally distributed continuous variables summarized as medians with interquartile ranges and compared using Mann-Whitney U tests, normally distributed continuous variables summarized as mean and standard deviation (SD), and compared using independent samples t-test. Categorical variables appeared as frequencies with percentages, analyzed by chi-square or Fisher’s exact tests as appropriate. To identify predictors associated with 30-day mortality while minimizing overfitting risk, we performed least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation. The optimal tuning parameter λ was selected using the one-standard-error rule (λ-1SE), which yields the most parsimonious model within one standard error of the minimum deviance. Variables with nonzero coefficients at λ-1SE were retained for further analysis. Variables selected by LASSO were entered into a multivariable logistic regression model to identify independent mortality predictors. To evaluate the combined predictive value of MHR and MELD, multivariable logistic regression analysis was performed with 30-day mortality as the dependent variable. MHR and MELD were entered as independent variables to construct the combined model. The predicted probability derived from this model was used as the combined predictor for subsequent analyses. Given the limited number of events (n = 17), we performed internal validation using bootstrap resampling with 1000 iterations. Optimism-corrected area under the curve (AUC) values were calculated to assess model stability. Diagnostic performance was evaluated through receiver operator characteristic (ROC) curve analysis, with optimal thresholds determined by Youden’s index and corresponding sensitivity-specificity pairs reported. Statistical significance was defined as two-tailed p-values below 0.05.

Results

Patient demographics and baseline characteristics

This study included a total of 244 middle-aged and elderly patients with HBV-DeCi. Of these patients, 227 (93.0%) survived for 30 days, while 17 (7.0%) did not. The main reasons for admission were ascites, hepatic encephalopathy and jaundice. Active infection was present in 104 (42.6%) patients at admission and 4 patients met the diagnostic criteria for Acute-on-Chronic Liver Failure (ACLF). A comparative analysis of baseline characteristics revealed significant differences in multiple laboratory parameters and clinical features between the two groups. Non-survivors exhibited significantly higher median levels of MHR (2.62 vs. 0.43, p < 0.001), WBC (6.8 × 10⁹/L vs. 3.9 × 10⁹/L, p < 0.001), MO (0.77 × 10⁹/L vs. 0.32 × 10⁹/L, p < 0.001), NE (5.34 × 10⁹/L vs. 2.39 × 10⁹/L, p < 0.001), PT (18.3 S vs. 15.5 S, p = 0.001), INR (1.60 vs. 1.36, p < 0.001), APTT (45 S vs. 33 S, p = 0.003) and TT (21.7 S vs. 19.5 S, p = 0.005). Compared to survivors, non-survivors had significantly higher median scores for ALT (48 U/L vs. 27 U/L, p = 0.012), AST (78 U/L vs. 39 U/L, p = 0.003), DBIL (147µmol/L vs. 11µmol/L, p < 0.001), UREA (8.4 mmol/L vs. 5.7 mmol/L, p = 0.006) and CREA (94µmol/L vs. 64µmol/L, p < 0.001). Furthermore, non-survivors had significantly higher median MELD scores (22 vs. 9, p < 0.001), APRI scores (3.6 vs. 1.8, p < 0.001), and FIB-4 scores (12 vs. 7, p = 0.001). Conversely, HDL levels (0.32 mmol/L vs. 0.78 mmol/L, p < 0.001), LDL levels (1.62 mmol/L vs. 1.83 mmol/L, p = 0.042) and ALB levels (28.3 g/L vs. 31.4 g/L, p = 0.025) were significantly lower in the non-survivor group (Table 1). Subgroup analysis based on age revealed that there is no significant difference in MHR between middle-aged (n = 191 patients) (0.43 (0.255–0.848) versus elderly (n = 55 patients) (0.546 (0.286 − 0.1029) (p = 0.190).

Table 1.

Baseline characteristics of patients with HBV-DeCi, stratified by 30-day mortality status

Characteristic	Overall N = 244	Survivors N = 227	Nonsurvivors N = 17	Statistic (U/T/χ2)	p-value
Gender(male), n (%)	173 (70.9%)	160 (70.5%)	13 (76.5%)	0.275	0.784
Age(years), M (Q1, Q3)	55 (49, 62)	54 (49, 62)	61 (50, 68)	-0.852	0.395
MHR, M (Q1, Q3)	0.45 (0.26, 0.91)	0.43 (0.24, 0.77)	2.62 (1.20, 4.92)	-5.070	< 0.001
WBC (×109/L), M (Q1, Q3)	4.0 (2.9, 6.3)	3.9 (2.8, 5.9)	6.8 (5.7, 7.9)	-3.541	< 0.001
LY (×109/L, M (Q1, Q3)	0.85 (0.54, 1.42)	0.84 (0.53, 1.40)	1.11 (0.61, 1.69)	-0.942	0.347
MO (×109/L), M (Q1, Q3)	0.34 (0.23, 0.58)	0.32 (0.22, 0.53)	0.77 (0.53, 1.04)	-4.041	< 0.001
NE (×109/L), M (Q1, Q3)	2.55 (1.68, 4.24)	2.39 (1.65, 4.12)	5.34 (3.53, 6.53)	-3.301	< 0.001
HGB (g/L), Mean ± SD	101 ± 25	101 ± 25	94 ± 27	0.967	0.346
PLT (×109/L), M (Q1, Q3)	63 (41, 102)	63 (41, 101)	63 (36, 104)	0.449	0.655
PT(S), M (Q1, Q3)	15.7 (14.3, 17.6)	15.5 (14.2, 17.3)	18.3 (15.5, 22.7)	-3.210	0.001
INR, M (Q1, Q3)	1.37 (1.25, 1.55)	1.36 (1.24, 1.53)	1.60 (1.37, 1.95)	-3.294	< 0.001
APTT (S), M (Q1, Q3)	34 (29, 41)	33 (29, 40)	45 (36, 60)	-3.014	0.003
FIB (g/L), M (Q1, Q3)	1.49 (1.12, 1.97)	1.50 (1.12, 1.97)	1.29 (1.02, 1.73)	-1.062	0.289
TT (S), M (Q1, Q3)	19.6 (18.3, 21.5)	19.5 (18.2, 21.2)	21.7 (20.9, 25.0)	-2.836	0.005
D-D (mg/L), M (Q1, Q3)	1.95 (0.75, 4.04)	1.88 (0.73, 4.07)	2.52 (1.81, 4.01)	-1.589	0.112
ALT (U/L), M (Q1, Q3)	27 (18, 57)	27 (18, 53)	48 (25, 117)	-2.512	0.012
AST (U/L), M (Q1, Q3)	41 (28, 74)	39 (28, 71)	78 (48, 178)	-2.989	0.003
DBIL (µmol/L), M (Q1, Q3)	12 (7, 26)	11 (7, 21)	147 (38, 266)	-5.038	< 0.001
HDL (mmol/L), M (Q1, Q3)	0.77 (0.50, 1.02)	0.78 (0.56, 1.03)	0.32 (0.21, 0.49)	-4.514	< 0.001
LDL (mmol/L), M (Q1, Q3)	1.79 (1.39, 2.32)	1.83 (1.40, 2.32)	1.62 (1.06, 1.95)	-2.038	0.042
ALB (g/L), Mean ± SD	31.2 ± 6.0	31.4 ± 6.0	28.3 ± 5.1	2.421	0.025
UREA (mmol/L), M (Q1, Q3)	5.8 (4.3, 8.1)	5.7 (4.3, 7.8)	8.4 (7.0, 14.6)	-2.733	0.006
CREA (µmol/L), M (Q1, Q3)	65 (53, 80)	64 (53, 76)	94 (74, 140)	-3.429	< 0.001
Na (mmol/L), M (Q1, Q3)	138.1 (136.1, 140.8)	138.1 (136.1, 140.9)	138.0 (132.9, 139.5)	-1.460	0.145
MELD, M (Q1, Q3)	9 (6, 14)	9 (6, 12)	22 (17, 26)	-5.435	< 0.001
APRI, M (Q1, Q3)	1.9 (1.0, 3.5)	1.8 (1.0, 3.3)	3.6 (2.2, 7.5)	-3.397	< 0.001
FIB-4 (g/L), M (Q1, Q3)	8 (5, 11)	7 (5, 11)	12 (8, 20)	-3.187	0.001

Data with normal distribution: Mean ± SD; Data with non-normal distribution: M (Q1, Q3). M: Median, N: number, Q1: 1st Quartile, Q3: 3st Quartile, SD: Standard deviation; ALB: Albumin; ALT: Alanine Aminotransferase; APTT: Activated Partial Thromboplastin Time; APRI: AST to Platelet Ratio Index; FIB-4: Fibrosis-4 Index; AST: Aspartate Aminotransferase; CREA: Creatinine; DBIL: Direct Bilirubin; D-D: D-Dimer; FIB: Fibrinogen; HB: Hemoglobin; HDL: High-Density Lipoprotein Cholesterol; INR: Prothrombin Time-International Normalized Ratio; LDL: Low-Density Lipoprotein Cholesterol; LY: Lymphocyte; MELD: Model for End-Stage Liver Disease; MHR: monocyte to high-density lipoprotein cholesterol ratio; MO: Monocyte; Na: Sodium; NE: Neutrophils; PLT: Platelet; PT: Prothrombin Time; TT: Thrombin Time; WBC: leukocyte

Factors associated with mortality

LASSO regression identified MHR, INR and DBIL as key predictors (Fig. 1). Subsequent multivariate analysis, including MHR, INR and DBIL, confirmed both MHR (Odds Ratio (OR) = 1.862, 95% CI: 1.354–2.561, p < 0.001) and DBIL (OR = 1.008, 95% CI: 1.003–1.013, p = 0.001) as independent factors of 30-day mortality in this HBV-DeCi cohort (Table 2).

Fig. 1.

Screening of variables based on LASSO regression. A: Cross-validation curve. The binomial deviance is plotted against log(λ). The left dashed vertical line indicates the λ value that minimizes the binomial deviance (λ-min), and the right dotted vertical line corresponds to the largest λ within one standard error of the minimum (λ-1SE). The λ-1SE was selected for the final model. B: Regularization path plot. Each colored line represents the coefficient trajectory of a variable as a function of log(λ). The dashed vertical line indicates the λ value selected using the one-standard-error rule (λ-1SE)

Table 2.

Multivariable logistic regression analysis of factors associated with 30-day mortality

Variables	Multivariate analysis
	OR	95% CI	p value
MHR	1.862	1.354–2.561	< 0.001
DBIL	1.008	1.003–1.013	0.001

DBIL: Direct Bilirubin; OR: Odds ratio; MHR: Monocyte to high-density lipoprotein cholesterol ratio

ROC curve analysis

Assessment of prognostic accuracy via ROC analysis yielded AUC values of 0.869 for MHR alone and 0.895 for the MELD score. The combined MELD + MHR model achieved an AUC of 0.917 (95% CI: 0.875–0.948), which was numerically higher than that of MHR alone (AUC = 0.869) and MELD alone (AUC = 0.895). However, DeLong test revealed that the improvement in AUC was not statistically significant compared with MELD alone (p = 0.169) or MHR alone (p = 0.164). Nevertheless, the combined model demonstrated the highest AUC among the three candidate predictors and showed favorable calibration and clinical utility in decision curve analysis. At an optimal cutoff of 0.073, the combined index delivered balanced performance with 82.4% sensitivity alongside 86.8% specificity (Table 3; Fig. 2). Bootstrap resampling (1000 iterations) for internal validation and the optimism-corrected C-index was 0.911. Kaplan-Meier analysis further confirmed the efficacy of the combined index in identifying high-risk patients (> 0.073) (Fig. 3).

Table 3.

Discriminative performance of MHR, MELD, and the combined model for predicting 30-day mortality

Variable	AUC	95% CI	Cut-off value	Sensitivity (%)	Specificity (%)	P value (vs. reference)
MHR	0.869	0.820–0.908	1.08	82.4	85.9	0.169
MELD	0.895	0.850–0.931	12.60	94.1	76.2	0.164
MELD + MHR	0.917	0.875–0.948	0.073	82.4	86.8	Reference

AUC: Area Under the Curve; MELD: Model for End-Stage Liver Disease; MHR: monocyte to high-density lipoprotein cholesterol ratio. Pairwise comparisons of AUCs (DeLong test)

Fig. 2.

Receiver operating characteristic curves of the prognostic performances of MHR、and MELD score for prediction of poor outcomes in HBV-DeCi patients

Fig. 3.

Kaplan-Meier analysis for the combined index in identifying high-risk patients

Discussion

Our analyses consistently demonstrated markedly elevated MHR levels among non-surviving patients compared to survivors. Multivariate analysis confirmed MHR’s independent prognostic value for 30-day mortality, with substantial predictive capacity demonstrated by ROC analysis. Notably, combining MHR with the conventional MELD score yielded an AUC of 0.917, representing a meaningful improvement over either marker alone. These findings suggest that MHR may serve as a promising composite biomarker for risk stratification in middle-aged and elderly HBV-DeCi patients.

Current evidence indicates that decompensated cirrhosis drives persistent innate immune activation, promoting both monocyte proliferation and functional enhancement [1–12]. These activated monocytes subsequently amplify hepatic inflammation through abundant release of pro-inflammatory mediators including TNF-α and IL-6 [13–14]. Meanwhile, progressive liver impairment substantially compromises HDL synthesis—a process critically dependent on hepatic function [15]. In the context of decompensated cirrhosis and ACLF, a reduced baseline hepatic reserve and longstanding circulatory dysfunction mean that even a lower volume of hepatocyte death can trigger the release of damage-associated molecular patterns and incite a potent systemic inflammatory response. Elevated MHR values thus quantify the prevailing imbalance between pro-inflammatory forces and protective HDL-mediated mechanisms, a disequilibrium closely associated with multi-organ dysfunction and mortality risk [16]. Our observations align perfectly with this mechanistic framework: non-survivors consistently demonstrated expanded monocyte populations, depressed HDL concentrations, and consequently elevated MHR measurements.

When compared with existing prognostic markers such as NLR, NHR, APRI, and FIB-4, MHR offers several potential advantages. Unlike APRI and FIB-4, which rely on transaminase and platelet values that may fluctuate with acute decompensation, MHR captures the integrated inflammatory-metabolic state. Moreover, while NLR and NHR reflect neutrophil-mediated inflammation, MHR specifically quantifies monocyte-driven immune responses coupled with the loss of HDL-mediated anti-inflammatory capacity—a pathway more directly implicated in cirrhosis progression and ACLF pathophysiology. This mechanistic specificity may explain the strong predictive performance observed in our study. These results coincide with growing recognition of MHR’s prognostic utility across medical specialties. Cardiovascular research has firmly established this ratio as a significant predictor of clinical outcomes in coronary artery disease [17], while hepatology reports have linked MHR to non-alcoholic steatohepatitis severity [18–19]. What distinguishes the present investigation is its specific focus on middle-aged and elderly HBV-DeCi patients. Of note, MELD demonstrated superior sensitivity in our cohort; thus, MHR may serve as a complementary tool to MELD rather than a replacement, potentially identifying patients at high risk due to inflammatory burden before severe organ dysfunction is fully captured by MELD. The significant enhancement in predictive power achieved by combining these two markers underscores their complementary value.

From a clinical implementation standpoint, MHR offers exceptional practicality [20–25]. Its component measures, monocyte counts and HDL values, represent routine, inexpensive laboratory parameters available in virtually all hospital settings. For individuals displaying elevated MHR values, such cases may benefit from intensified monitoring and multimodal management approaches.

Several methodological constraints merit consideration when interpreting these findings. The retrospective single-center design, while providing initial insights, necessitates confirmation through prospective multicenter investigations with larger cohorts. A particularly important limitation is the small number of mortality events, which increases the risk of model overfitting and limits the complexity of analyses that could be performed. To mitigate this risk, we employed LASSO regression for variable selection, maintained a parsimonious modeling approach, and performed bootstrap internal validation. Our relatively limited sample size introduces potential statistical instability that should be addressed in future validation studies. Furthermore, the exclusive focus on 30-day mortality leaves unanswered questions regarding MHR’s prognostic utility over extended timeframes, as 90-day analysis was constrained by sample size. Additionally, the absence of external validation precludes definitive conclusions regarding the generalizability of our findings. Future research initiatives would benefit from prospective designs across multiple centers with expanded sample sizes and extended follow-up to verify and extend these observations.

In conclusion, in this middle-aged and elderly cohort with HBV-DeCi, we have demonstrated that MHR serves as a potential predictor of 30-day mortality. The integration of MHR with the conventional MELD score creates a promising prognostic tool that may enhance risk stratification accuracy for middle-aged and elderly patients. This readily accessible composite model has the potential to help clinicians to identify high-risk individuals more effectively, prompting timely interventions and optimized management strategies tailored to this vulnerable population. However, these findings should be considered preliminary, and external validation in larger, prospective multicenter cohorts is warranted before clinical implementation.

Author contributions

S B C, Y B W and F J H contributed equally to this work. J Z and B P designed the study. S B C and Y B W wrote the manuscript. J Z was responsible for data collection, analysis, and interpretation. All the authors have revised, read and approved the final manuscript.

Funding

This research was supported by the Suqian Sci&Tech Program (KY202310) and Suqian Natural Science Foundation Project (K202423).

Data availability

The data could be obtained from the corresponding author.

Declarations

Ethical approval

The study was fulfilled in line with the principles presented in the Declaration of Helsinki, and was approved by the local ethics committee of the First Affiliated Hospital with Nanjing Medical University (2023-SR-022). The institutional Ethics Committee of the First Affiliated Hospital with Nanjing Medical University approved the research protocol and granted a waiver for individual informed consent given the retrospective design and no identifiable data were used.

Consent for publication

Not applicable.

Disclosure

None.

Competing interests

The authors declare no competing interests.