Lymphocyte‐to‐Monocyte Ratio as a Predictor of Recompensation in Patients With Decompensated Primary Biliary Cholangitis

ABSTRACT

Background

Primary biliary cholangitis (PBC) patients in the decompensated stage face poor prognoses, with recompensation being crucial for improving long‐term outcomes.

Aim

This study aims to evaluate the predictive value of the lymphocyte‐to‐monocyte ratio (LMR) for recompensation in decompensated PBC patients.

Methods

We retrospectively analysed 410 patients with PBC‐related decompensated cirrhosis receiving ursodeoxycholic acid (UDCA) treatment. The association between the LMR and recompensation was examined using Cox regression analysis, with additional trend analysis performed based on LMR quartiles. The predictive accuracy of the LMR, neutrophil‐to‐lymphocyte ratio (NLR), systemic immune‐inflammation index (SII) and platelet‐to‐lymphocyte ratio (PLR) was evaluated using receiver operating characteristic (ROC) curve analysis. Sensitivity analyses were conducted to confirm the robustness of the findings.

Results

During follow‐up, among 401 patients with decompensated cirrhosis (age: 60.0 [IQR: 53.0–69.0] years; 88.3% female), 105 patients (26.18%) achieved recompensation. Multivariate Cox regression analysis showed that higher LMR was an independent promoting factor for recompensation after adjusting for all confounding factors in the model (HR = 1.415, 95% CI: 1.264–1.585, p < 0.001), with a linear positive correlation trend. ROC curve analysis demonstrated that LMR had superior predictive performance compared to other inflammatory markers (SII, PLR, NLR), with an area under the curve (AUC) of 0.787 (95% CI: 0.736–0.838, p < 0.001).

Conclusion

LMR serves as a robust independent predictor for recompensation in decompensated PBC patients.

This study identifies lymphocyte‐to‐monocyte ratio as a robust independent predictor for recompensation in decompensated primary biliary cholangitis patients, demonstrating superior predictive performance compared to other inflammatory markers and an accessible prognostic tool for clinical management.

1. Introduction

Primary biliary cholangitis (PBC) is a chronic inflammatory autoimmune cholestatic liver disease characterised by portal inflammation and immune‐mediated destruction of intrahepatic bile ducts. The disease predominantly affects middle‐aged women, with approximately 95% of patients testing positive for the specific serum marker, anti‐mitochondrial antibody (AMA) [1, 2]. PBC follows a chronic progressive course, eventually leading to cirrhosis, liver failure and even death. Epidemiological data indicate that about 15% of patients are already in the advanced stages of the disease at the time of initial diagnosis [3], and the cumulative incidence of cirrhosis within 10 years reaches 40% [4]. Patients with cirrhosis are at significantly increased risk of decompensating events, such as ascites, hepatic encephalopathy and oesophageal variceal bleeding. Once decompensation occurs, the transplant‐free survival rate of these patients declines markedly [5]. However, clinical observations have shown that some decompensated PBC patients can achieve ‘recompensation’—a return to a compensated state of liver function—following treatment with ursodeoxycholic acid (UDCA) and other therapies, which significantly improves their long‐term survival [6]. Therefore, early identification of patients who are likely to achieve recompensation and timely intervention are crucial for improving the prognosis of PBC patients.

In recent years, the role of inflammation in the pathogenesis of PBC has garnered increasing attention. Numerous studies have demonstrated that peripheral blood inflammatory markers are associated with disease activity, severity, and prognosis in PBC [7, 8]. The lymphocyte‐to‐monocyte ratio (LMR) is a non‐invasive indicator that integrates immune dysregulation and inflammatory responses. It has advantages such as ease of calculation and accessibility, and it has shown significant value in prognostic evaluations for malignancies and cardiovascular diseases [9, 10, 11]. However, the predictive role of LMR in decompensated PBC patients remains unclear.

This study aims to investigate, for the first time, the relationship between LMR and the occurrence of recompensation in decompensated PBC patients, and to validate its predictive performance and robustness. The findings are expected to provide new insights into prognostic evaluation for decompensated PBC patients.

2. Methods

2.1. Study Population

This was a single‐centre, retrospective cohort study that included PBC patients with decompensation who visited the Second Affiliated Hospital of Kunming Medical University between January 2013 and December 2023. Patients were initially screened using International Classification of Diseases (ICD) codes, followed by manual chart review according to the predefined inclusion/exclusion criteria (Figure 1).

FIGURE 1.

Flowchart of participant selection.

Inclusion criteria were:

1. Diagnosis of PBC based on the European Association for the Study of the Liver (EASL) guidelines.

2. Initial diagnosis of decompensated PBC, with decompensating events defined as ascites, variceal bleeding, or hepatic encephalopathy.

Exclusion criteria were:

1. Coexisting viral hepatitis, alcoholic liver disease, or autoimmune hepatitis‐primary biliary cholangitis (AIH‐PBC) overlap syndrome.

2. Coexisting malignancies or infections (presence of infection records, antibiotic administration and laboratory or imaging findings indicative of infection).

3. Patients with UDCA treatment duration of less than 1 year or without any UDCA treatment.

4. Insufficient follow‐up data.

5. Patients receiving corticosteroids or immunosuppressants.

This retrospective cohort study was approved by the Ethics Committee of the Second Affiliated Hospital of Kunming Medical University and conducted in accordance with the Declaration of Helsinki (YJ‐2023‐96).

2.2. Definition of Recompensation

The definition of recompensation is based on the Baveno VII consensus [12] and EASL clinical practice guidelines [2]. Recompensation was defined as:

1. Successful etiological treatment: UDCA treatment response (ALP < 1.67 ULN after 12 months).

2. Resolution of decompensating events: ascites (cessation of diuretics), encephalopathy (cessation of lactulose/rifaximin) and no recurrence of variceal bleeding (for at least 12 months).

3. Stable improvement in liver function: albumin > 35 g/L, international normalised ratio (INR) < 1.50 and total bilirubin < 34 μmol/L.

2.3. Data Collection

We retrospectively reviewed patients' electronic medical records and collected demographic characteristics, timing and type of the first decompensating event, laboratory parameters, inflammatory markers and medical history at the time of initial decompensation. The corrected BMI was calculated using the ‘dry weight’ method (for patients with mild, moderate, or severe ascites, current weight was reduced by 5%, 10%, or 15% respectively; for patients with peripheral edema, an additional 5% was reduced) [13]. Systemic immune‐inflammation biomarkers, including systemic immune‐inflammation index (SII), neutrophil‐to‐lymphocyte ratio (NLR), platelet‐to‐lymphocyte ratio (PLR) and LMR, were calculated as follows:

SII = platelet count × neutrophil count/lymphocyte count;

NLR = neutrophil count/lymphocyte count;

PLR = platelet count/lymphocyte count;

LMR = lymphocyte count/monocyte count.

2.4. Follow‐Up and Outcomes

Patient follow‐up was conducted through both outpatient visits and inpatient records. For decompensated patients, monthly biochemical monitoring was conducted, including blood cell analysis, liver function (ALT, AST, ALP, bilirubin), renal function and coagulation function; liver ultrasound was performed every 6 months; and gastroscopy was conducted annually to screen for oesophagogastric varices. The primary endpoint of this study was the achievement of recompensation. Follow‐up duration was defined as the time interval from the first decompensating event to either recompensation, death, or loss to follow‐up.

2.5. Statistical Analysis

Statistical analyses were performed using SPSS 30.0 and R 4.2.0 software. Continuous variables with non‐normal distributions were expressed as medians (interquartile ranges, IQR) and compared between groups using the Mann–Whitney U test. Categorical data were reported as percentages and analysed using the chi‐squared test (χ²) or Fisher's exact test. Kaplan–Meier survival curves were constructed, and group differences in survival were compared using the Log‐rank test. To determine the independent predictive significance of LMR for recompensation in decompensated PBC patients, Cox proportional hazards model analysis was performed, and multiple models were established to adjust for potential confounding factors stepwise. The predictive accuracy for recompensation in decompensated patients during the entire follow‐up period was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC) and 95% confidence intervals (CI). The AUCs of predictive models were compared using the DeLong test. A two‐sided p < 0.05 was considered statistically significant. Missing data was minimal (< 6%), confined to three antibody tests (anti‐M2: 3.2%, anti‐gp210/sp100: 5.7%). Complete‐case analysis was used as all demographic, outcome, and key laboratory variables had no missing values.

3. Results

3.1. Baseline Characteristics of the Study Population

A total of 401 decompensated PBC patients were included in this study (Figure 1). Of these, 349 patients (87.03%) were diagnosed through positive AMA or specific antinuclear antibodies (anti‐sp100 or anti‐gp210), while 52 patients (12.97%) were diagnosed by liver biopsy. The median age of the patients was 60 years (IQR: 53.0–69.0 years), and 343 (85.54%) were female. Among the initial decompensating events, ascites was the most common manifestation (250 cases, 62.34%), followed by variceal bleeding (98 cases, 24.44%) and overt hepatic encephalopathy (53 cases, 13.22%). Some patients presented with comorbid conditions, including hypertension (70 cases, 17.46%), coronary artery disease (11 cases, 2.74%) and diabetes mellitus (45 cases, 11.22%). Detailed baseline characteristics of the patients are presented in Table 1.

TABLE 1.

Baseline characteristics of the patients with decompensated PBC.

Characteristic	Overall (n = 401)	No recompensation (n = 296)	Recompensation (n = 105)	p
Age (years), Median (IQR)	60 (53–69)	61 (53–69)	58 (52–68.75)	0.186
Sex (female)	334 (83.29%)	239 (80.74%)	95 (90.48%)	0.022
Adjusted BMI (Kg/m2) a	20.17 (17.78–22.57)	19.48 (17.4–21.89)	21.78 (19.57–24.2)	< 0.001
Hypertension	70 (17.46%)	47 (15.88%)	23 (21.9%)	0.162
Coronary artery disease (%)	11 (2.74%)	6 (2.03%)	5 (4.76%)	0.165
Diabetes mellitus (%)	45 (11.22%)	38 (12.84%)	7 (6.67%)	0.085
Decompensation event
Ascites	303 (75.9%)	245 (83.3%)	58 (55.2%)	< 0.001
Variceal haemorrhage	134 (33.5%)	105 (35.6%)	29 (27.6%)	0.138
Hepatic encephalopathy	53 (13.22%)	47 (15.9%)	6 (5.7%)	0.008
Multiple (≥ 2) decompensation events	123 (30.7%)	103 (34.8%)	20 (19%)	0.003
TBil (μmol/L), Median (IQR)	29.8 (16.83–70.4)	38.1 (22.15–106.5)	16.45 (13.63–24.63)	< 0.001
TBA (μmol/L), Median (IQR)	46.9 (17.13–129.5)	64.4 (23.68–171)	20.85 (8.55–46.6)	< 0.001
ALT (×ULN), Median (IQR)	40 (23.25–71.75)	42 (26–74)	32 (20–64.75)	0.015
AST (×ULN), Median (IQR)	65 (38.25–98)	72 (42.25–106.75)	47 (31–72)	< 0.001
ALP (×ULN), Median (IQR)	196.5 (128–330.75)	216 (134.75–370)	157 (121.25–241)	< 0.001
GGT (×ULN), Median (IQR)	125 (60–298)	133 (63.25–296.75)	96 (48–305.75)	0.109
Platelets (×109/l)	113 (70–176)	105.5 (69–160.75)	142 (79–206)	< 0.001
INR	1.15 (1.01–1.31)	1.19 (1.06–1.38)	1.01 (0.94–1.14)	< 0.001
AMA‐M2 antibodies	317 (81.7%)	241 (83.97%)	76 (75.25%)	0.072
SP100	55 (14.55)	10 (10.1%)	45 (16.13%)	0.184
GP210	138 (36.51%)	23 (23.23%)	115 (41.22%)	0.002
SII	292.95 (149.58–510.19)	299.03 (156.02–517.51)	273.42 (140.31–495.73)	0.259
PLR	110.53 (73.28–160.7)	114.04 (75.68–165.3)	95 (66.23–151.42)	0.035
NLR	2.45 (1.7–4.03)	2.69 (1.86–4.55)	2.06 (1.49–2.76)	< 0.001
LMR	3.29 (2.2–4.55)	2.93 (2–3.96)	4.74 (3.49–6.51)	< 0.001

Note: Data presented as number n (%) or median (IQR). p‐Values in bold indicate statistical significance (p < 0.05). Missing data are highlighted in superscript.

Abbreviations: ALP, alkaline phosphatase; ALT, alanine aminotransferase; AMA‐M2, anti‐mitochondrial antibody M2; Anti‐gp210 antibody, Anti‐glycoprotein 210 antibody; Anti‐sp100 antibody, Anti‐sp100 nuclear antigen antibody; AST, aspartate aminotransferase; BMI, body mass index; GGT, gamma‐glutamyltransferase; INR, international normalised ratio; LMR, lymphocyte‐to‐monocyte ratio; NLR, neutrophil‐to‐lymphocyte ratio; PLR, platelet‐to‐lymphocyte ratio; SII, systemic immune‐inflammation index; TBA, total bile acid; TBil, total bilirubin.

^aAdjusted BMI was calculated using the ‘dry weight’ method (for patients with mild, moderate, or severe ascites, current weight was reduced by 5%, 10%, or 15% respectively; for patients with peripheral edema, an additional 5% was reduced) [13].

3.2. Occurrence of Recompensation and Its Relationship With Prognosis

During the follow‐up period, 105 patients (26.18%) achieved recompensation (Figure 2). The cumulative survival rates at 1, 3 and 5 years in the recompensation group were 96.97%, 92.23% and 86.98%, respectively, which were significantly higher than those observed in the non‐recompensation group (78.04%, 58.85% and 45.77%, respectively; p < 0.001, Figure 3). When patients were stratified by quartiles of LMR, those in the highest quartile group (Q4) demonstrated a significantly higher recompensation rate compared to patients in the lowest quartile group (Q1) (p < 0.001, Figure 4).

FIGURE 2.

Natural history and recompensation outcomes stratified by LMR quartiles in patients with decompensated PBC.

FIGURE 3.

Kaplan–Meier survival curves for patients with re‐compensation and no re‐compensation.

FIGURE 4.

Cumulative incidence of re‐compensation based on LMR quartiles.

3.3. Cox Regression Analysis of LMR as an Independent Predictor of Recompensation

Univariate Cox regression analysis revealed that increased levels of TBIL, TBA, AST, ALP and anti‐gp210 antibody positivity were unfavourable factors for recompensation, whereas higher BMI and LMR were favourable factors (p < 0.05). Multivariate Cox regression analysis was conducted using three stepwise adjustment models to evaluate the association between LMR and recompensation: Model A (adjusted for age, sex and BMI): BMI (HR: 1.076, 95% CI: 1.029–1.124, p = 0.001) and LMR (HR: 1.355, 95% CI: 1.242–1.479, p < 0.001) were positively associated with recompensation. Model B (Model A plus adjustment for comorbidities): BMI (HR: 1.072, 95% CI: 1.025–1.122, p = 0.002) and LMR (HR: 1.351, 95% CI: 1.235–1.478, p < 0.001) remained significant predictors. Model C (Model B plus adjustment for multiple (≥ 2) decompensation events and biochemical parameters): LMR (HR: 1.273, 95% CI: 1.156–1.401, p < 0.001), BMI (HR: 1.059, 95% CI: 1.004–1.116, p = 0.036) and total bilirubin (HR: 0.986, 95% CI: 0.974–0.999, p = 0.037) were identified as independent predictors of recompensation (Table 2).

TABLE 2.

Cox regression analysis of LMR as an independent predictor of re‐compensation.

Variable	Univariate analysis		Multivariate analysis
			Model A		Model B		Model C
	HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p
Age	0.994 (0.977–1.011)	0.469	0.997 (0.979–1.014)	0.705	0.995 (0.976–1.013)	0.573	0.985 (0.966–1.005)	0.149
Sex (male/female)	1.927 (1.004–3.698)	0.049	1.377 (0.711–2.668)	0.343	1.327 (0.683–2.578)	0.404	1.021 (0.507–2.058)	0.953
Adjusted BMI (Kg/m2) a	1.102 (1.058–1.148)	< 0.001	1.076 (1.029–1.124)	0.001	1.072 (1.025–1.122)	0.002	1.059 (1.004–1.116)	0.036
Hypertension	1.340 (0.844–2.128)	0.215			1.289 (0.782–2.124)	0.320	1.321 (0.778–2.245)	0.303
Coronary artery disease	2.118 (0.862–5.203)	0.102			1.343 (0.500–3.610)	0.558	1.657 (0.552–4.972)	0.367
Diabetes mellitus	0.600 (0.279–1.291)	0.191			0.730 (0.330–1.616)	0.437	0.805 (0.355–1.826)	0.604
Multiple (≥ 2) decompensation events	0.420 (0.170–1.039)	0.060					0.957 (0.547–1.673)	0.877
TBil (μmol/L)	0.979 (0.969–0.989)	< 0.001					0.986 (0.974–0.999)	0.037
TBA (μmol/L)	0.993 (0.989–0.996)	< 0.001					0.997 (0.993–1.002)	0.228
ALT (μ/l)	1.000 (0.997–1.002)	0.824					0.999 (0.993–1.005)	0.672
AST (μ/l)	0.996 (0.991–1.000)	0.033					1.002 (0.994–1.011)	0.568
ALP (μ/l)	0.998 (0.997–1.000)	0.006					0.999 (0.997–1.000)	0.144
GGT (μ/l)	1.000 (0.999–1.000)	0.193					1.000 (0.999–1.001)	0.668
Platelets (×109/l)	1.002 (1.000–1.003)	0.081					1.002 (1.000–1.005)	0.090
AMA‐M2 antibodies	0.649 (0.413–1.019)	0.060					0.833 (0.510–1.361)	0.467
Anti‐sp100 antibody	0.737 (0.383–1.417)	0.360					0.763 (0.386–1.507)	0.436
Anti‐gp210 antibody	0.488 (0.306–0.779)	0.003					0.658 (0.400–1.081)	0.098
LMR	1.398 (1.285–1.521)	< 0.001	1.355 (1.242–1.479)	< 0.001	1.351 (1.235–1.478)	< 0.001	1.273 (1.156–1.401)	< 0.001

Note: Model A: Adjusted for age, sex, adjusted BMI and LMR. Model B: Adjusted for age, sex, adjusted BMI, hypertension, coronary artery disease, diabetes mellitus and LMR. Model C: Adjusted for age, sex, adjusted BMI, hypertension, coronary artery disease, diabetes mellitus, multiple (≥ 2) decompensation events, total bilirubin, total bile acids, ALT, AST, ALP, GGT, anti‐sp100 antibody, anti‐gp210 antibody and LMR.

Abbreviations: ALP, alkaline phosphatase; ALT, alanine aminotransferase; AMA‐M2, anti‐mitochondrial antibody M2; anti‐gp210 antibody, anti‐glycoprotein 210 antibody; Anti‐sp100 antibody, anti‐sp100 nuclear antigen antibody; AST, aspartate aminotransferase; BMI, body mass index; GGT, gamma‐glutamyltransferase; LMR, lymphocyte‐to‐monocyte ratio; TBA, total bile acid; TBil, total bilirubin.

^aAdjusted BMI was calculated using the ‘dry weight’ method (for patients with mild, moderate, or severe ascites, current weight was reduced by 5%, 10%, or 15% respectively; for patients with peripheral edema, an additional 5% was reduced) [13].

3.4. Trend Analysis and Nonlinear Examination

Using LMR quartiles (Q1–Q4), univariate Cox regression analysis demonstrated that, compared to the lowest quartile (Q1), the risk of recompensation was significantly higher in the highest quartile group (Q4) (HR: 5.978, 95% CI: 2.845–12.561, p < 0.001). After adjusting for various confounding factors, Models A, B and C all confirmed a positive association between LMR levels and recompensation risk (Model A: HR: 4.947, 95% CI: 2.325–10.527; Model B: HR: 4.700, 95% CI: 2.197–10.056; Model C: HR: 4.925, 95% CI: 2.041–11.884; all p < 0.001) (Table 3). Restricted cubic spline analysis revealed a linear positive relationship between LMR and recompensation (p for nonlinearity = 0.662, Figure 5), corroborating the findings from the regression analyses.

TABLE 3.

Trend analysis of re‐compensation in decompensated PBC patients based on LMR.

Variable	Univariate analysis		Multivariate analysis
			Model A		Model B		Model C
	HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p
LMR (quartile)
Quantile 1	Reference	< 0.001	Reference	< 0.001	Reference	< 0.001	Reference	< 0.001
Quantile 2	1.308 (0.534–3.2)	0.557	1.200 (0.487–2.958)	0.692	1.140 (0.461–2.817)	0.776	1.695 (0.614–4.681)	0.309
Quantile 3	3.376 (1.547–7.365)	0.002	2.951 (1.345–6.473)	0.007	2.913 (1.324–6.409)	0.008	3.356 (1.361–8.275)	0.009
Quantile 4	5.978 (2.845–12.561)	< 0.001	4.947 (2.325–10.527)	< 0.001	4.700 (2.197–10.056)	< 0.001	4.925 (2.041–11.884)	< 0.001
p for trend	1.924 (1.565–2.366)	< 0.001	1.819 (1.475–2.244)	< 0.001	1.798 (1.455–2.223)	< 0.001	1.680 (1.334–2.115)	< 0.001

Note: Model A: Adjusted for age, sex, adjusted BMI and LMR. Model B: Adjusted for age, sex, adjusted BMI, hypertension, coronary artery disease, diabetes mellitus and LMR. Model C: Adjusted for age, sex, adjusted BMI, hypertension, coronary artery disease, diabetes mellitus, multiple (≥ 2) decompensation events, total bilirubin, total bile acids, ALT, AST, ALP, GGT, anti‐sp100 antibody, anti‐gp210 antibody and LMR.

FIGURE 5.

Restricted cubic spline (RCS) analysis for the nonlinear association between LMR and the risk of re‐compensation.

3.5. Predictive Performance of Inflammatory Markers

The predictive capacity of SII, PLR, NLR and LMR for recompensation in PBC patients was evaluated using ROC curve analysis. The results demonstrated that LMR exhibited the highest predictive performance, with an area under the curve (AUC) of 0.787 (95% CI: 0.736–0.838, p < 0.001), significantly superior to NLR (AUC = 0.658, 95% CI: 0.601–0.715, p < 0.001), PLR (AUC = 0.569, 95% CI: 0.508–0.631, p = 0.035) and SII (AUC = 0.537, 95% CI: 0.475–0.599, p = 0.259) (Figure 6). Statistical comparison of the AUCs using the DeLong test revealed significant differences between LMR and all other inflammatory markers (all p < 0.001), confirming the superior discriminative ability of LMR. NLR demonstrated significantly better performance compared to SII (p < 0.001) and PLR (p < 0.01), while no significant difference was observed between SII and PLR (p > 0.05) (Figure 7). Further time‐dependent ROC analysis revealed that the AUCs of LMR at 1, 3 and 5 years were 0.742 (95% CI: 0.676–0.809), 0.742 (95% CI: 0.677–0.807) and 0.748 (95% CI: 0.669–0.827), respectively (p < 0.001), indicating consistent and robust prognostic capability over time (Figure 8).

FIGURE 6.

ROC curves of inflammatory markers for predicting re‐compensation in PBC patients.

FIGURE 7.

Statistical comparison of AUC values among inflammatory markers (LMR, NLR, SII and PLR) using DeLong test.

FIGURE 8.

Time‐dependent ROC curve of LMR for predicting re‐compensation in PBC patients.

3.6. Sensitivity Analysis

To evaluate the potential confounding effects of comorbidities, we performed a sensitivity analysis by excluding patients with hypertension, coronary artery disease and diabetes mellitus. The subsequent Cox regression analysis confirmed that LMR (HR = 1.265, 95% CI: 1.135–1.410, p < 0.001) remained a significant independent predictor of recompensation, thus validating the robustness of our primary findings (Table S1).

4. Discussion

This study is the first to investigate the predictive value of LMR for the occurrence of recompensation in patients with decompensated PBC. Through a retrospective cohort study involving 410 patients with decompensated PBC, we identified LMR as an independent predictor of recompensation, with higher LMR levels significantly associated with an increased likelihood of achieving recompensation. Further analyses using ROC curves and time‐dependent ROC analyses confirmed the high predictive performance and dynamic stability of LMR. Additionally, sensitivity analyses demonstrated that the predictive power of LMR was not affected by comorbid conditions such as hypertension, coronary artery disease, or diabetes mellitus, thus validating the robustness of our findings.

LMR reflects the proportion of lymphocytes to monocytes in peripheral blood, providing a comprehensive assessment of the patient's immune‐inflammatory status. Our findings demonstrate that higher LMR values are significantly associated with stronger compensatory capacity in PBC patients, which is consistent with previous research suggesting elevated LMR predicts favourable prognosis in PBC patients [7]. Specifically, bile duct injury originates from immune interactions between biliary epithelial cells and surrounding inflammatory cells [14]. Studies have shown that CD4+ lymphocytes constitute the primary inflammatory cells surrounding damaged bile ducts [15, 16], and disease progression and recovery are closely related to the balance between these CD4+ T cell subsets. During disease progression, CD4+ T cell subset imbalance manifests as a marked predominance of pro‐inflammatory components (Th1/Th17), while regulatory T cells (Tregs) decrease in number or become functionally impaired [17, 18], leading to disruption of immune tolerance. Conversely, during successful hepatic compensation, this imbalance reverses, promoting restoration of immune tolerance, alleviating peribiliary inflammation and ultimately facilitating biliary epithelial repair and preservation of functional bile ducts. Monocytes play a ‘dual‐damage’ role in PBC, primarily accelerating disease progression through pro‐inflammatory responses and hepatic fibrosis. They secrete pro‐inflammatory cytokines such as IL‐12 and IL‐23 [19], driving Th1 and Th17 differentiation [20, 21] and amplifying local inflammatory cascades. More critically, monocytes promote hepatic fibrosis through IL‐17‐mediated hepatic stellate cell activation [22, 23], directly accelerating the development of portal hypertension. In our study cohort, portal hypertension‐related complications (ascites 62.34%, variceal bleeding 24.44%) constituted the vast majority (86.78%) of initial decompensation events. Decreased monocyte counts suggest attenuated pro‐fibrotic activity, alleviating portal pressure and reducing the risk of associated complications, which explains why patients with higher LMR demonstrate stronger recompensation capacity. Notably, significant synergistic interactions exist between monocytes and lymphocytes, jointly constructing the immune‐inflammatory network in PBC. Monocytes promote Th1 and Th17 cell differentiation by secreting cytokines such as TNF‐α and IL‐6, while these activated T cells in turn produce IFN‐γ, further stimulating monocyte activation, forming a self‐sustaining inflammatory cycle. During recompensation, elevated LMR reflects not only changes in individual cell counts but also a critical shift in cell population proportions toward lymphocyte predominance. This reconstruction of immune balance indicates that protective lymphocyte responses exceed pro‐inflammatory monocyte activity. In this process, UDCA treatment further promotes immune tolerance restoration, alleviates peribiliary inflammation, and ultimately facilitates biliary epithelial repair by inhibiting the pro‐inflammatory cascade. It is worth mentioning that cirrhotic patients face significantly increased infection risk, and the ability of elevated LMR to predict recompensation may relate to enhanced anti‐infection capacity. Although our study strictly excluded patients with active infection, higher LMR reflects a more balanced immune status, providing stronger immune defence and reducing secondary infection risk, thereby increasing recompensation probability.

Compared with other commonly used inflammatory markers, NLR primarily reflects neutrophil‐mediated acute inflammatory responses; PLR focuses on the interrelationship between platelet activation and immune function; while SII integrates multiple parameters to provide a comprehensive assessment of systemic immune‐inflammatory status. The excellent predictive capability demonstrated by LMR in our study likely stems from its specificity in reflecting the unique immune‐inflammatory balance in PBC patients, precisely capturing cellular immune changes closely related to disease progression and recovery [14, 15, 16, 17, 18, 19, 20, 21, 22, 23].

Furthermore, our univariate Cox regression analysis revealed that elevated total bilirubin, total bile acids, AST, ALP levels and anti‐gp210 antibody positivity negatively impact recompensation, consistent with pathophysiological mechanisms of cholestasis and hepatocellular injury in PBC. Additionally, we found that higher BMI positively correlates with recompensation in PBC patients (HR = 1.059, 95% CI: 1.005–1.116, p = 0.032), possibly reflecting better nutritional status and energy reserves in decompensated patients, providing necessary foundations for hepatic repair and regeneration. The key finding of this study is the significant positive correlation between LMR and recompensation, where higher LMR levels were associated with a greater likelihood of achieving recompensation. This relationship remained robust even after adjusting for potential confounders. Although LMR has been shown to have strong predictive value in other diseases such as cancer and cardiovascular diseases [9, 10, 11], its application in PBC remains limited.

To our knowledge, this study is the first to focus specifically on decompensated PBC patients.

Currently, no single biomarker exists to predict the likelihood of recompensation in decompensated PBC patients. This study proposes LMR as a relatively effective predictive marker. LMR is an economical and readily accessible inflammatory marker that can be quickly obtained through routine blood tests without additional cost. This could help clinicians identify patients with high potential for recompensation and develop more individualised treatment strategies. Additionally, higher LMR levels could boost the confidence of PBC patients in facing the long disease course. However, our study has several limitations. First, as a single‐centre retrospective analysis, unmeasured confounders may persist despite statistical adjustments, necessitating validation of the LMR‐recompensation association in future multicentre prospective cohorts to improve generalisability. Second, focusing on peripheral blood markers, we did not investigate the intrahepatic immune microenvironment or cytokine mechanisms critical to PBC pathogenesis. Exploring these biological pathways will aid clinical translation of LMR in PBC management. Lastly, we did not compare LMR with established risk scores (UK‐PBC, GLOBE) due to their mortality/transplant focus and potential overlap with recompensation‐defining variables. Future studies could assess LMR's complementary role in broader prognostic frameworks for PBC.

5. Conclusion

In summary, this study demonstrates for the first time that LMR is an independent predictor of recompensation in patients with decompensated PBC, with higher LMR levels significantly increasing the likelihood of achieving recompensation. Moreover, the predictive performance of LMR surpasses that of other established inflammatory markers, with stable dynamic monitoring capabilities. As an economical and readily available haematological marker, LMR provides a novel approach for evaluating prognosis and developing individualised treatment plans for decompensated PBC patients. However, further validation in larger‐scale, multicentre prospective studies and in‐depth exploration of its underlying biological mechanisms are needed to promote the clinical translation and widespread application of LMR in PBC management.

Author Contributions

Huiying Lin: writing – original draft, investigation, conceptualization, methodology, validation, visualization, software, formal analysis, data curation, resources. Mengyao Zheng: conceptualization, methodology, funding acquisition, supervision, project administration, writing – review and editing. Yaqin Huang: data curation. Huilin Zhu: data curation. Lili Zhang: data curation. Wenting Yang: data curation. Hongjin Wang: data curation. Tingting Yin: data curation, software, investigation. Min Zhou: data curation, investigation. Hongtao Lei: writing – review and editing. Jinhui Yang: writing – review and editing, funding acquisition, supervision, project administration. Wenlin Tai: writing – review and editing.

Ethics Statement

The present study protocol was approved by the Ethics Committee of the Second Affiliated Hospital of Kunming Medical University. The requirement for informed consent was waived by the Ethics Committee of the Second Affiliated Hospital of Kunming Medical University due to the retrospective nature of the study.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to ethical restrictions.