Introduction
Bacterial infection (BI) is common in patients with cirrhosis and is the leading cause of hospital admission.1 The incidence of BI in cirrhotic patients is as high as 34% per year,2,3 and about one-third of those admitted to the hospital develop a hospital-acquired infection.3 BI, which is the most common precondition for acute-on-chronic liver failure (ACLF), can lead to further decompensation, including oesophageal and gastric varices bleeding (EGVB), hepatorenal syndrome, and hepatic encephalopathy.4 Therefore, the early diagnosis of BI and the identification of associated complications, such as ACLF, are crucial for individualized treatment to improve the poor prognosis of cirrhotic patients with BI.
Early recognition of BI is challenging.5,6 Cirrhotic patients may not exhibit typical signs of infection, such as fever, in the early stages of infection.7 Routine serum biomarkers have limited diagnostic performance in cirrhotic patients.2,8 The white blood cell (WBC) count may be affected by hypersplenism8 and cirrhosis-associated immune dysfunction (CAID).9 C-reactive protein (CRP) may serve as an indirect marker of acute inflammatory responses.10 In cases of renal dysfunction or comorbid conditions, elevated procalcitonin levels may be misinterpreted as a sign of cirrhosis.11,12 Therefore, there is a need for new diagnostic biomarkers for BI.
The poor prognosis of cirrhotic patients with BI may be linked to immune dysfunction.13 Adaptive immune deficiency, partly due to suppressed or insufficient T-cell responses,14 could be a significant risk factor for immune paralysis and bacterial susceptibility in cirrhotic patients.15 The impaired antimicrobial response in conditions like acute liver failure and alcoholic acute hepatitis has been associated with CD4+ and CD8+ T cells.16–18
Mitochondrial metabolic regulation is essential for the transformation of T cells from naïve to active states, and from active to memory cells.19,20 Chen et al.21 concluded that the mitochondrial function of CD8+ T cells is essential for anti-tumor immunity. Mitochondria regulate T-cell maturation, differentiation, and function in various ways.22 Mitochondrial function plays an important role in regulating T-cell immune metabolism.23 The mitochondrial membrane potential (MMP), the potential difference across the mitochondrial inner membrane, reflects the real-time metabolic state of cells.24 The MMP is involved in the production of ATP and reactive oxygen species (ROS):25 the higher the MMP, the greater the ATP production rate.26,27 An abnormal MMP can significantly impair the function and survival of immune cells.24 Mitochondrial mass (MM) refers to the total amount of protein available in the mitochondrial respiratory chain, representing the upper limit of a cell’s metabolic capacity.28
We aimed to evaluate the diagnostic efficiency of MM and MMP in T cells from peripheral blood (PB) for predicting BI and ACLF within 90 days after hospitalization in cirrhotic patients.
Methods
Patient selection
A prospective observational cohort study (TBNK COHORT STUDY) of cirrhotic inpatients was conducted at the Second Hospital of Nanjing from January to November 2023. This cohort study evaluated changes in the mitochondrial function of T cells, B cells, and NK cells in PB in the progression of cirrhosis and their clinical application. Additionally, we collected PB from patients for cytokine assays to assess whether mitochondria-related indicators reflect systemic inflammatory changes in CAIDs from March 2024 to June 2024. This study was part of the TBNK Clinical Programme.
The inclusion criteria were patients aged ≥18 and ≤80 years with cirrhosis, based on histological, biochemical, ultrasonic, or radiological findings. Both patients with compensated and decompensated cirrhosis were included in the study. The exclusion criteria were as follows: (1) ACLF present at admission; (2) acute infection within the previous week or prophylactic antibiotics within one week; (3) human immunodeficiency virus infection; (4) treatment with corticosteroids, immunosuppressants, or cytotoxic drugs; (5) hepatocellular carcinoma and concomitant non-liver-related malignant tumors; (6) previous liver lobectomy or liver transplantation; (7) severe heart failure (New York Heart Association class 4), chronic obstructive pulmonary disease (Global Initiative for Chronic Obstructive Lung Disease grade 3–4), or kidney dysfunction (chronic kidney disease level 5) diagnosed before cirrhosis; (8) pregnancy or lactation; and (9) missing follow-up data.
Study design
Upon admission, patients were evaluated for inclusion and exclusion criteria. Patients meeting the criteria were introduced to the program, and those who agreed to participate signed informed consent. Patients underwent routine physical examinations, laboratory tests, and imaging exams (ultrasound and CT). Three to five milliliters of peripheral blood was collected from patients on the day of admission or the day after admission (depending on the time of admission) and transported to the laboratory for testing within 2 h. Demographic, clinical, laboratory, imaging, and treatment data were collected (antibiotic use was recorded in detail). Baseline indicators in the Lis-System were recorded on the day of admission or the day after admission.
The detection time of blood samples was considered day 0. Follow-up was performed at one week, one month, and three months. Phone and WeChat follow-up were conducted as needed. During follow-up, any condition requiring active management was treated accordingly.
The study endpoints were BI and ACLF.
Bacterial infections were diagnosed according to the following criteria: (1) spontaneous bacterial peritonitis: ascites multinucleate cell count ≥ 250/mm3; (2) urinary tract infection: at least one of the following signs or symptoms (fever ≥ 38°C, urgency to urinate, frequent urination, dysuria, or suprapubic tenderness) and positive urine culture, or at least two signs or symptoms (fever ≥ 38°C, urgency to urinate, frequent urination, dysuria, or suprapubic tenderness) with urine WBC greater than 10/µL; (3) pneumonia: radiographic examination suggesting new or progressive osmotic or cavernous changes, plus at least one of the following: fever ≥ 38°C, WBC ≥ 12,000/mm3 or ≤4,000/mm3, and at least one of the following: new purulent sputum or change in sputum character, new cough, dyspnea or shortness of breath (20 breaths per minute), rales or bronchial breathing sounds, or worsening gas exchange, and/or positive culture from blood, pleural effusion, or specimens obtained from the trachea, aspiration, bronchoalveolar lavage, or biopsy; and (4) acute biliary tract infection: radiographic examination (CT or MRCP) showing exudative or necrotic changes in the biliary tract, plus at least one of the following: fever ≥ 38°C, WBC ≥ 12,000/mm3 or ≤4,000/mm3, plus local or systemic inflammatory changes (abdominal pain, jaundice, positive Murphy sign, Charcot triad); (5) spontaneous bacteraemia: in the absence of a known infection source, at least one of the following: fever ≥ 38°C, chills, and hypotension, with positive blood cultures (bilateral and 2-culturing-bottle) and (6) the upper respiratory tract infection: acute pharyngitis, tonsillitis, laryngitis, sinusitis, otitis media, with fever and WBC ≥ 12,000/mm3 or ≤ 4,000/mm3, with or without clear etiological evidence, and significant improvement after antibiotic treatment (according to ‘Guideline for primary care of acute upper respiratory tract infection (2018)’, DOI: 10.3760/cma.j.issn.1671-7368.2019.05.005). BI at other sites were diagnosed according to the corresponding diagnostic criteria.
ACLF was diagnosed following the EASL guidelines.29
Two physicians regularly reviewed the data to detect errors and assess the accuracy of the collected data. The researchers who tested the cell counts, MM, and MMP were blinded to the clinical status of the patients. After verifying the collected clinical variables, statistical analysis was performed.
Measurement of MM and MMP, mitochondrial ultrastructure observation, autophagy-related protein detection, and cytokine concentration determination
This process is described in detail in the Supplementary File 1.
Ethics
The research adhered to both the Declaration of Helsinki and the Declaration of Istanbul and was approved by the Ethics Committee of the Second Hospital of Nanjing (2023-LY-ky-022). All patients provided written informed consent.
Statistical analysis
Clinical indexes with missing values of less than 10% were imputed using multiple imputation methods.
The Survival and Survminer packages in R software were used for Kaplan–Meier curves and log-rank statistical calculations. The rms package was used to draw restricted cubic splines, and the ggpubr package was used to draw fitted curves.
Data are presented as frequencies and percentages, means and standard deviations, or medians and interquartile ranges, as appropriate. Baseline characteristics were compared using Fisher’s exact test for categorical variables, Student’s t-test for continuous variables, and the Wilcoxon rank-sum test for ordinal and continuous variables. Fitted curves were drawn to assess the relationships between cell counts or markers (inflammation, immune exhaustion, and liver disease) and mitochondria-related indicators, and Spearman correlation coefficients were calculated. The receiver operating characteristic (ROC) curves for BI and ACLF at 90 days were plotted, and the areas under the ROC curves (AUROC) were calculated. Considering the possibility of a nonlinear correlation, the optimal cut-off for MM or MMP of T cells was determined by plotting a 5-knot restricted cubic spline (RCS). C-indexes were also calculated to compare the diagnostic efficacy of different indicators. Kaplan–Meier curves based on mitochondria-related indicators are presented. The association between risk factors and the rates of BI and ACLF was assessed using univariable and multivariable Cox proportional hazards regression models, and forward regression (Wald) analysis was employed to select variables for inclusion in the multivariable model. Time-dependent ROC curves were used to evaluate the diagnostic performance of the conventional model combined with T-cell mitochondria-related indicators. The adjusted RCS was plotted for sensitivity analysis. Statistical analyses were performed using GraphPad software (version 9.0), the Statistical Package for Social Sciences (version 22.0), RStudio (version 4.2.1), and Python (version 3.10).
The level of significance was set at the two-sided 5% level.
Results
Cohort baseline characteristics
According to the inclusion and exclusion criteria, 235 cirrhotic patients were analyzed (mean age 52.65 ± 12.93 years, 118 males, 120 in the decompensated stage) (Supplementary Fig. 1), 45 of whom were newly diagnosed with liver disease. The baseline characteristics of the patients are shown in Table 1.
Table 1Baseline characteristics of the full cohort
| Patients with compensated cirrhosis (N = 115) | Patients with decompensated cirrhosis (N = 120) | P-value |
|---|
| Basic characteristics | | | |
| Age (years) | 48.50 ± 12.67 | 56.63 ± 11.94 | <0.001 |
| Sex (males/females) | 49/66 | 69/51 | 0.02 |
| Aetiology (AIH/PBC-AIH overlapping syndrome/PBC/Alcoholic /HCV/HBV/MASH) | 19/15/18/4/10/35/14 | 19/8/11/10/10/61/1 | <0.001 |
| Complications | | | |
| Ascites (yes/no) | 0/115 | 105/15 | <0.001 |
| EGV (yes/no) | 1/114 | 56/64 | <0.001 |
| HE (yes/no) | 0/115 | 26/94 | <0.001 |
| Splenectomy (yes/no) | 1/114 | 20/100 | <0.001 |
| TIPS (yes/no) | 0/115 | 4/116 | 0.048 |
| Diabetes (yes/no) | 18/97 | 27/93 | 0.18 |
| Cardiovascular disease (yes/no) | 36/79 | 33/87 | 0.52 |
| chronic kidney diseases (yes/no) | 1/114 | 6/114 | 0.06 |
| Chronic lung disease (yes/no) | 3/112 | 1/119 | 0.29 |
| Aetiological treatment | | | |
| Receiving the aetiological treatment (yes/no) | 92/23 | 87/33 | 0.17 |
| Aetiological control or Biochemical response* (yes/no) | 26/89 | 39/81 | 0.09 |
| Clinical characteristics | | | |
| Higher than ULN of high sensitivity CRP (yes/no) | 10/105 | 41/79 | <0.001 |
| Higher than ULN of CRP (yes/no) | 6/109 | 15/105 | 0.05 |
| White blood cells (*109/L) | 5.48 ± 2.03 | 4.50 ± 1.98 | <0.001 |
| Neutrophils percentage (%) | 55.18 ± 12.37 | 57.29 ± 14.86 | 0.24 |
| Haemoglobin (g/L) | 130.19 ± 25.11 | 116.15 ± 25.47 | <0.001 |
| Platelet (*1012/L) | 172.82 ± 75.21 | 93.56 ± 60.78 | <0.001 |
| Prothrombin time (s) | 11.51 ± 1.41 | 15.22 ± 5.25 | <0.001 |
| International normalized ratio | 1.04 ± 0.13 | 1.42 ± 0.52 | <0.001 |
| Total bilirubin (µmol/L) | 13.00 (8.80, 29.30) | 27.60(17.40, 59.80) | <0.001 |
| Albumin (g/L) | 41.08 ± 6.85 | 36.84 ± 15.32 | 0.007 |
| Globulin (g/L) | 27.24 ± 5.45 | 30.44 ± 7.45 | <0.001 |
| Alanine transaminase (U/L) | 79.60 (35.40, 182.60) | 28.10 (19.20, 50.30) | <0.001 |
| Aspartate aminotransferase (U/L) | 44.30 (29.30, 117.10) | 36.70 (24.00, 75.60) | 0.08 |
| Gamma-glutamyl transpeptidase (U/L) | 79.70 (36.60, 171.20) | 50.00 (21.40, 92.40) | <0.001 |
| Alkaline phosphatase (U/L) | 93.00 (57.00,155.00) | 123.00 (90.00, 164.00) | 0.18 |
| Creatinine (µmol/L) | 63.55 ± 15.27 | 72.6 ± 37.00 | 0.02 |
| Blood urea nitrogen (mmol/L) | 4.53 ± 1.24 | 5.59 ± 4.95 | 0.02 |
| Serum sodium (mmol/L) | 138.78 ± 10.08 | 139.58 ± 3.91 | 0.42 |
| C3 (g/L) | 1.13 ± 0.67 | 0.79 ± 0.32 | <0.001 |
| C4 (g/L) | 0.54 ± 0.43 | 0.15 ± 0.10 | 0.21 |
| IgA (g/L) | 2.50 ± 1.17 | 4.19 ± 2.71 | <0.001 |
| IgG (g/L) | 14.95 ± 4.80 | 18.41 ± 6.75 | <0.001 |
| IgM (g/L) | 2.01 ± 1.93 | 2.17 ± 1.76 | 0.53 |
| Liver stiffness (kPa) | 16.34 ± 5.71 | 31.82 ± 22.59 | <0.001 |
| Immune-related index | | | |
| Absolute value of CD45+ cells (/µL) | 576.18 ± 232.68 | 427.00 ± 228.96 | <0.001 |
| Percentage of CD45+ cells (%) | 39.04 ± 11.92 | 32.68 ± 13.61 | <0.001 |
| Absolute value of CD3+ cells (/µL) | 1,538.26 ± 640.40 | 947.10 ± 585.53 | <0.001 |
| Percentage of CD3+ cells (%) | 75.00 ± 8.54 | 67.39 ± 12.89 | <0.001 |
| Absolute value of CD3+CD4+ cells (/µL) | 895.74 ± 426.47 | 556.13 ± 365.36 | <0.001 |
| Percentage of CD3+CD4+ cells (%) | 43.42 ± 8.87 | 39.83 ± 12.08 | 0.01 |
| Absolute value of CD3+CD8+ cells (/µL) | 5,167.86 ± 260.02 | 311.26 ± 232.13 | <0.001 |
| Percentage of CD3+CD8+ cells (%) | 25.42 ± 8.99 | 21.69 ± 9.28 | 0.002 |
| Absolute value of CD3-CD19+ cells (/µL) | 227.14 ± 177.02 | 202.14 ± 198.22 | 0.31 |
| Percentage of CD3-CD19+ cells (%) | 10.87 ± 6.09 | 13.88 ± 9.51 | 0.004 |
| Absolute value of CD3-CD56+ cells (/µL) | 244.80 ± 206.98 | 245.93 ± 243.40 | 0.97 |
| Percentage of CD3-CD56+ cells (%) | 11.90 ± 6.63 | 15.92 ± 10.22 | <0.001 |
| Ratio of CD4 positive cells to CD8 positive cells | 2.03 ± 1.12 | 2.25 ± 1.51 | 0.21 |
| MM of CD3+ cells | 2.04 ± 1.38 | 2.58 ± 1.55 | 0.005 |
| MM of CD3+CD4+ cells | 2.74 ± 1.54 | 3.30 ± 1.90 | 0.01 |
| MM of CD3+CD8+ cells | 1.02 ± 0.81 | 1.44 ± 1.23 | 0.002 |
| MM of CD3+CD4+CD8+ cells | 2.52 ± 1.84 | 3.17 ± 2.18 | 0.02 |
| MM of CD3+CD4−CD8− cells | 0.56 ± 0.43 | 0.75 ± 0.90 | 0.04 |
| MM of CD3-CD19+ cells | 1.01 ± 0.75 | 1.15 ± 1.07 | 0.23 |
| MM of CD3-CD56+ cells | 0.80 ± 0.79 | 1.03 ± 1.38 | 0.12 |
| MMP of CD3+ cells (%) | 37.64 ± 11.70 | 32.55 ± 11.73 | 0.001 |
| MMP of CD3+CD4+ cells (%) | 24.75 ± 13.13 | 20.97 ± 13.54 | 0.03 |
| MMP of CD3+CD8+ cells (%) | 54.30 ± 11.91 | 46.15 ± 12.82 | <0.001 |
| MMP of CD3+CD4+CD8+ cells (%) | 33.96 ± 17.37 | 29.17 ± 17.37 | 0.03 |
| MMP of CD3+CD4-CD8- cells (%) | 62.74 ± 15.49 | 57.52 ± 19.27 | 0.02 |
| MMP of CD3-CD19+ cells (%) | 50.47 ± 16.51 | 48.94 ± 17.01 | 0.48 |
| MMP of CD3-CD56+ cells (%) | 60.03 ± 16.63 | 56.16 ± 19.98 | 0.11 |
Mitochondria-related indicators to predict infection within 90 days
BI occurred in 97 patients (41.28%) during follow-up, 32 (9.79%) of whom had BI during their hospital stay, 10 (4.25%) of whom were readmitted, and the remaining patients received outpatient or emergency treatment with oral or intravenous antibiotics, nutritional support, etc. The most common sites of infection were the upper respiratory tract (52, 53.61%) and urinary tract (20, 20.62%). The distribution of infection sites is shown in Supplementary Figure 2. Notably, the 90-day incidence of BI was significantly different between decompensated and compensated patients (hazard ratio (HR) 5.41, 95% confidence interval (CI): 3.62–8.08, P < 0.001, Supplementary Fig. 3), as was the distribution of causes (P = 0.009, Supplementary Fig. 2).
Compared with patients who did not develop BI within 90 days, those who did had fewer CD3+ T cells, CD4+ T cells, and CD8+ T cells (all P < 0.001, Supplementary Table 1A). These results suggested that the immune function of patients with cirrhosis was significantly impaired and that those at high risk for infection had more severe immune deficiencies. There were significant differences in mitochondria-related indicators of T cells between patients who developed and did not develop BI within 90 days (Supplementary Table 1A).
Previous studies have suggested that T-cell mitochondrial function is closely related to T-cell fate.19 We generated a fitting curve to assess the correlation between T-cell count and MM or MMP. MM did not correlate with cell count for T cells, CD4+ T cells, or CD8+ T cells (P > 0.05), while MMP was weakly correlated with cell count (Supplementary Fig. 4).
The AUROCs of MM and MMP for predicting BI within 90 days were calculated. The MMP of CD8+ T cells demonstrated the best predictive performance (AUROC = 0.79, 95% CI 0.73–0.85, P < 0.001) (Table 2A).
Table 2Performance of the mitochondria-related indicators in predicting BI within 90 days
| Mitochondria-related indicators | AUROC (95% CI) | P-value |
|---|
| (A) Performance of the mitochondria-related indicators in predicting BI within 90 days in the total cohort |
| MM of CD3+ cells | 0.65 (0.58, 0.72) | <0.001 |
| MM of CD3+CD4+ cells | 0.63 (0.56, 0.71) | <0.001 |
| MM of CD3+CD8+ cells | 0.72 (0.65,0.78) | <0.001 |
| MM of CD3+CD4+CD8+ cells | 0.59 (0.52, 0.67) | 0.01 |
| MM of CD3+CD4−CD8− cells | 0.64 (0.56, 0.71) | <0.001 |
| MMP of CD3+ cells | 0.65 (0.58, 0.72) | <0.001 |
| MMP of CD3+CD4+ cells | 0.62 (0.54, 0.69) | 0.002 |
| MMP of CD3+CD8+ cells | 0.79 (0.73, 0.85) | <0.001 |
| MMP of CD3+CD4+CD8+ cells | 0.58 (0.50, 0.65) | 0.049 |
| MMP of CD3+CD4−CD8− cells | 0.64 (0.57, 0.71) | <0.001 |
| (B) Performance of the mitochondria-related indicators in predicting BI within 90 days in compensated patients |
| MM of CD3+ cells | 0.56 (0.42, 0.70) | 0.43 |
| MM of CD3+CD4+ cells | 0.53 (0.40, 0.67) | 0.66 |
| MM of CD3+CD8+ cells | 0.62 (0.48,0.78) | 0.09 |
| MM of CD3+CD4+CD8+ cells | 0.63 (0.50, 0.76) | 0.08 |
| MM of CD3+CD4−CD8− cells | 0.60 (0.46, 0.74) | 0.18 |
| MMP of CD3+ cells | 0.65 (0.51, 0.78) | 0.04 |
| MMP of CD3+CD4+ cells | 0.56 (0.41, 0.71) | 0.40 |
| MMP of CD3+CD8+ cells | 0.69 (0.56, 0.81) | 0.01 |
| MMP of CD3+CD4+CD8+ cells | 0.66 (0.54, 0.79) | 0.02 |
| MMP of CD3+CD4−CD8− cells | 0.72 (0.60, 0.83) | 0.003 |
| (C) Performance of the mitochondria-related indicators in predicting BI within 90 days in decompensated patients |
| MM of CD3+ cells | 0.66 (0.56, 0.76) | 0.005 |
| MM of CD3+CD4+ cells | 0.66 (0.55, 0.76) | 0.005 |
| MM of CD3+CD8+ cells | 0.75 (0.66,0.85) | <0.001 |
| MM of CD3+CD4+CD8+ cells | 0.53 (0.43, 0.64) | 0.54 |
| MM of CD3+CD4−CD8− cells | 0.68 (0.58, 0.78) | 0.001 |
| MMP of CD3+ cells | 0.60 (0.49, 0.71) | 0.06 |
| MMP of CD3+CD4+ cells | 0.59 (0.49, 0.70) | 0.09 |
| MMP of CD3+CD8+ cells | 0.81 (0.73, 0.85) | <0.001 |
| MMP of CD3+CD4+CD8+ cells | 0.52 (0.41, 0.62) | 0.76 |
| MMP of CD3+CD4−CD8− cells | 0.60 (0.49, 0.70) | 0.08 |
As shown in Supplementary Figure 5, we used RCS modeling and visualization to predict the relationship between MM or MMP of T cells and the risk of infection. There was a nonlinear relationship between MM and infection risk (nonlinear P < 0.05) (Supplementary Fig. 5). When the MM of CD3+ cells was less than 2.07, the risk of infection remained relatively stable, but above that threshold, it began to increase rapidly (nonlinear P < 0.001) (Supplementary Fig. 5A). When the number of CD8+ T cells was less than 0.86, the risk of infection remained stable, and above that value, it increased rapidly (nonlinear P < 0.05) (Supplementary Fig. 5E). In contrast, MMP had a linear relationship with the risk of infection: as MMP decreased, the risk of infection gradually increased. The cut-off values were 34.87 (the MMP of T cells), 19.66 (the MMP of CD4+ T cells), and 49.48 (the MMP of CD8+ T cells) (Supplementary Fig. 5B, D, F).
Mitochondria-related indicators to predict ACLF within 90 days
Twenty-nine (12.34%) patients developed ACLF. Consistent with the literature, BI (19, 68.97%) was the most common risk factor. Pulmonary infection (13, 44.83%) accounted for the majority, with iatrogenic and healthcare-associated infections being the most common pulmonary infections. There were significant differences in mitochondria-related indicators of T cells between patients who developed and did not develop ACLF within 90 days (Supplementary Table 1B).
The AUROCs of MM and MMP for predicting ACLF within 90 days were calculated, with the MMP of CD3+CD8+ cells showing superior performance (AUROC = 0.75, 95% CI 0.66–0.83, P < 0.001) (Table 3). Unexpectedly, the mitochondria-related indicators of CD4−CD8− T cells also predicted ACLF well (MM: AUROC = 0.74, 95% CI 0.64–0.83, P < 0.001; MMP: AUROC = 0.72, 95% CI 0.63–0.81, P < 0.001) (Table 3), even though they were not good predictors of BI.
Table 3Performance of mitochondria-related indicators for predicting ACLF
| Mitochondria-related indicators | AUROC (95% CI) | P-value |
|---|
| (A) Performance of the mitochondria-related indicators in predicting ACLF within 90 days in the total cohort |
| MM of CD3+ cells | 0.71(0.62, 0.80) | <0.001 |
| MM of CD3+CD4+ cells | 0.71 (0.61, 0.80) | <0.001 |
| MM of CD3+CD8+ cells | 0.69 (0.60,0.78) | <0.001 |
| MM of CD3+CD4+CD8+ cells | 0.54 (0.42, 0.66) | 0.49 |
| MM of CD3+CD4−CD8− cells | 0.74 (0.64, 0.83) | <0.001 |
| MMP of CD3+ cells | 0.66 (0.57, 0.75) | 0.005 |
| MMP of CD3+CD4+ cells | 0.60 (0.48, 0.71) | 0.10 |
| MMP of CD3+CD8+ cells | 0.75 (0.66, 0.83) | <0.001 |
| MMP of CD3+CD4+CD8+ cells | 0.51 (0.40, 0.63) | 0.80 |
| MMP of CD3+CD4−CD8− cells | 0.72 (0.63, 0.81) | <0.001 |
| (B) Performance of the mitochondria-related indicators in predicting ACLF within 90 days in decompensated patients |
| MM of CD3+ cells | 0.67 (0.56, 0.77) | 0.007 |
| MM of CD3+CD4+ cells | 0.67 (0.56, 0.78) | 0.007 |
| MM of CD3+CD8+ cells | 0.64 (0.54,0.75) | 0.02 |
| MM of CD3+CD4+CD8+ cells | 0.51 (0.38, 0.64) | 0.84 |
| MM of CD3+CD4−CD8− cells | 0.74 (0.63, 0.84) | <0.001 |
| MMP of CD3+ cells | 0.60 (0.49, 0.71) | 0.11 |
| MMP of CD3+CD4+ cells | 0.54 (0.42, 0.66) | 0.54 |
| MMP of CD3+CD8+ cells | 0.68 (0.57, 0.79) | 0.004 |
| MMP of CD3+CD4+CD8+ cells | 0.57 (0.45, 0.69) | 0.26 |
| MMP of CD3+CD4−CD8− cells | 0.70 (0.60, 0.80) | 0.001 |
As shown in Supplementary Figure 6, RCS modeling and visualization were used to predict the relationship between MM or MMP of T cells and the risk of ACLF. The MMP of CD8+ T cells had the strongest correlation with the risk of ACLF, and the threshold value was similar to that for BI prediction (49.48 to 49.45) (Supplementary Fig. 6).
Patients with low MMP in CD8+ T cells had a higher risk of BI and ACLF
We then performed a Cox analysis to predict BI. In the univariable analysis, 41 factors were included as candidate variables in the multivariable model (Table 4). The multivariable Cox regression model showed that the MMP of CD8+ T cells was an independent risk factor for BI (HR 0.96, 95% CI 0.94–0.98; P < 0.001; Table 4). We also performed a Cox analysis for predicting ACLF, which concluded that CRP, total bilirubin, international normalized ratio, creatine, CD8+ T cell counts, and the MMP of CD8+ T cells were independent risk factors (Table 5).
Table 4Cox regression evaluation of factors associated with BI
| Variable | Univariable analysis
| Multivariable analysis
|
|---|
| Hazard ratio | P-value | Hazard ratio | P-value |
|---|
| Age (y) | 1.03 (1.01, 1.04) | 0.002 | … | … |
| Stage (compensated/decompensated) | 5.29 (3.19, 8.75) | <0.001 | 1.92 (1.02. 3.60) | 0.04 |
| Splenectomy (no/yes) | 0.37 (0.22, 0.62) | <0.001 | … | … |
| Receiving the aetiological treatment (no/yes) | 0.56 (0.37, 0.86) | 0.008 | … | … |
| Ascites (no/yes) | 3.61 (2.32, 5.62) | <0.001 | … | … |
| EGV (no/yes) | 3.11 (2.08, 4.67) | <0.001 | 1.86 (1.15, 3.01) | 0.01 |
| HE (no/yes) | 5.01 (3.10, 8.09) | <0.001 | … | … |
| Higher than ULN of high sensitivity CRP (yes/no) | 7.33 (4.79, 11.22) | <0.001 | 3.01 (1.84, 4.93) | <0.001 |
| Higher than ULN of CRP (yes/no) | 5.60 (3.29, 9.52) | <0.001 | 2.25 (1.20, 4.20) | 0.01 |
| Neutrophils percentage (%) | 1.03 (1.01, 1.05) | 0.001 | … | … |
| Haemoglobin (g/L) | 0.98 (0.98, 0.99) | <0.001 | 0.99 (0.98, 1.00) | 0.44 |
| Platelet (*1012/L) | 0.99 (0.99, 0.99) | <0.001 | … | … |
| Prothrombin time (s) | 1.14 (1.11, 1.17) | <0.001 | 0.83 (0.73, 0.96) | 0.01 |
| International normalized ratio | 3.49 (2.67, 4.55) | <0.001 | 12.80 (3.188, 51.40) | <0.001 |
| Total bilirubin (µmol/L) | 1.01 (1.00, 1.01) | <0.001 | … | … |
| Albumin (g/L) | 0.92 (0.89, 0.95) | <0.001 | … | … |
| Globulin (g/L) | 1.05 (1.02, 1.08) | 0.002 | … | … |
| Aspartate aminotransferase (U/L) | 1.00 (1.00, 1.00) | 0.01 | … | … |
| Alkaline phosphatase (U/L) | 1.00 (1.00, 1.00) | 0.004 | 1.00 (1.00, 1.00) | 0.01 |
| C3 (g/L) | 0.13 (0.06, 0.27) | <0.001 | … | … |
| C4 (g/L) | 0.00 (0.00, 0.02) | <0.001 | … | … |
| IgA (g/L) | 1.20 (1.13, 1.27) | <0.001 | … | … |
| IgG (g/L) | 1.06 (1.03, 1.09) | <0.001 | … | … |
| IgM (g/L) | 1.10 (1.02, 1.12) | 0.02 | … | … |
| Absolute value of CD45+ cells (/µL) | 1.00 (1.00, 1.00) | <0.001 | … | … |
| Percentage of CD45+ cells (%) | 0.96 (0.94, 0.98) | <0.001 | … | … |
| Absolute value of CD3+ cells (/µL) | 1.00 (1.00, 1.00) | <0.001 | … | … |
| Percentage of CD3+ cells (%) | 0.97 (0.95, 0.98) | <0.001 | … | … |
| Absolute value of CD3+CD4+ cells (/µL) | 1.00 (1.00, 1.00) | <0.001 | … | … |
| Percentage of CD3+CD4+ cells (%) | 0.97 (0.95, 0.99) | 0.001 | … | … |
| Absolute value of CD3+CD8+ cells (/µL) | 1.00 (1.00, 1.00) | <0.001 | … | … |
| MM of CD3+ cells | 1.21 (1.08, 1.34) | <0.001 | … | … |
| MM of CD3+CD4+ cells | 1.16 (1.05, 1.28) | 0.003 | … | … |
| MM of CD3+CD8+ cells | 1.38 (1.21, 1.57) | <0.001 | … | … |
| MM of CD3+CD4+CD8+ cells | 1.11 (1.02, 1.21) | 0.02 | … | … |
| MM of CD3+CD4−CD8− cells | 1.31 (1.10, 1.56) | 0.002 | … | … |
| MMP of CD3+ cells (%) | 0.97 (0.95, 0.99) | <0.001 | … | … |
| MMP of CD3+CD4+ cells (%) | 0.98 (0.96, 1.00) | 0.02 | … | … |
| MMP of CD3+CD8+ cells (%) | 0.93 (0.92, 0.95) | <0.001 | 0.96 (0.94, 0.98) | <0.001 |
| MMP of CD3+CD4+CD8+ cells (%) | 0.99 (0.98, 1.00) | 0.11 | … | … |
| MMP of CD3+CD4−CD8− cells (%) | 0.98 (0.97, 0.99) | <0.001 | … | … |
Table 5Cox regression evaluation of factors associated with ACLF
| Variable | Univariable analysis
| Multivariable analysis
|
|---|
| Hazard ratio | P-value | Hazard eatio | P-value |
|---|
| Stage (compensated/decompensated) | 69.18 (3.48, 1,375.44) | 0.005 | … | … |
| Ascites (no/yes) | 16.86 (4.01, 70.93) | <0.001 | … | … |
| EGV (no/yes) | 3.07 (1.48, 6.35) | 0.003 | … | … |
| HE (no/yes) | 10.99 (5.27, 22.91) | <0.001 | … | … |
| Higher than ULN of high sensitivity CRP (yes/no) | 21.24 (8.08, 55.84) | <0.001 | 5.94 (2.14, 16.51) | <0.001 |
| Higher than ULN of CRP (yes/no) | 8.91 (4.23, 18.74) | <0.001 | … | … |
| Neutrophils percentage (%) | 1.07 (1.03, 1.10) | <0.001 | … | … |
| Haemoglobin (g/L) | 0.98 (0.97, 0.99) | <0.001 | … | … |
| Platelet (*1012/L) | 0.99 (0.98, 0.99) | <0.001 | … | … |
| Prothrombin time (s) | 1.23 (1.17, 1.28) | <0.001 | | |
| International normalized ratio | 6.03 (4.07, 8.93) | <0.001 | 4.36 (2.31, 8.22) | <0.001 |
| Total bilirubin (µmol/L) | 1.01 (1.01, 1.01) | <0.001 | 1.01 (1.00, 1.01) | <0.001 |
| Globulin (g/L) | 1.06 (1.01, 1.11) | 0.02 | … | … |
| Aspartate aminotransferase (U/L) | 1.00 (1.00, 1.00) | 0.03 | … | … |
| Creatinine (µmol/L) | 1.01 (1.00, 1.02) | 0.003 | 1.01 (1.01, 1.02) | 0.001 |
| C3 (g/L) | 0.02 (0.00, 0.06) | <0.001 | | |
| C4 (g/L) | 0.00 (0.00, 0.01) | <0.001 | … | … |
| IgA (g/L) | 1.24 (1.14, 1.35) | <0.001 | … | … |
| IgG (g/L) | 1.10 (1.06, 1.15) | <0.001 | … | … |
| Absolute value of CD45+ cells (/µL) | 1.00 (0.99, 1.00) | <0.001 | … | … |
| Percentage of CD45+ cells (%) | 0.94 (0.91, 0.97) | <0.001 | … | … |
| Absolute value of CD3+ cells (/µL) | 1.00 (1.00, 1.00) | <0.001 | … | … |
| Percentage of CD3+ cells (%) | 0.97 (0.94, 0.99) | 0.01 | … | … |
| Absolute value of CD3+CD4+ cells (/µL) | 1.00 (1.00, 1.00) | <0.001 | … | … |
| Absolute value of CD3+CD8+ cells (/µL) | 0.99 (0.99, 1.00) | <0.001 | 1.00 (0.99, 1.00) | 0.007 |
| MM of CD3+ cells | 1.28 (1.06, 1.54) | 0.009 | … | … |
| MM of CD3+CD4+ cells | 1.31 (1.12, 1.53) | 0.001 | … | … |
| MM of CD3+CD8+ cells | 1.30 (1.02, 1.66) | 0.04 | … | … |
| MM of CD3+CD4−CD8− cells | 1.51 (1.180, 1.93) | 0.001 | … | … |
| MMP of CD3+ cells (%) | 0.96 (0.92, 0.99) | 0.007 | … | … |
| MMP of CD3+CD8+ cells (%) | 0.93 (0.91, 0.96) | <0.001 | 0.94 (0.90, 0.97) | <0.001 |
| MMP of CD3+CD4−CD8− cells (%) | 0.96 (0.94, 0.98) | <0.001 | … | … |
Patients in the cohort were divided into two groups according to the cut-off value for the MMP of CD8+ T cells. The BI-free survival rate differed between patients with high and low MMP of CD8+ T cells (HR 4.09, 95% CI 2.73–6.14; P < 0.001; C-index: 0.68 ± 0.02; Fig. 1A). The ACLF-free survival rate also differed between patients with high and low MMP of CD8+ T cells (HR 4.47, 95% CI 2.15–9.27; P < 0.001; C-index: 0.67 ± 0.04; Fig. 1D).
The diagnostic efficiency of MMP of CD8+ T cells was superior to traditional indices
We also calculated C-indices to compare the diagnostic efficacy of different indicators. In predicting the occurrence of BI within 90 days, the MMP of CD8+ T cells, as a single index, demonstrated better diagnostic efficacy than traditional indices, including absolute T cell count, high sensitivity CRP, CRP, WBC, and the percentage of neutrophils (C-index: 0.75 vs 0.70, 0.69, 0.58, 0.55, and 0.63, Supplementary Table 2A). In predicting ACLF, however, the diagnostic efficacy of the single index, the MMP of CD8+ T cells, was inferior to high-sensitivity CRP, total T cell counts, and the percentage of neutrophils (Supplementary Table 2A).
Time-dependent ROC curves were then drawn to evaluate the diagnostic efficacy of combining MMP with traditional models. The MMP of CD8+ T cells improved the ability of Child-Turcotte-Pugh (CTP) score and model for end-stage liver disease (MELD) score to predict BI (AUROC at 90 days: CTP: 0.81 to 0.89; MELD: 0.78 to 0.89; Supplementary Fig. 7). The predictive abilities of the CTP score and MELD score for ACLF could also be strengthened by combining them with the MMP of CD8+ T cells (AUROC at 90 days: CTP: 0.94 to 0.96; MELD: 0.95 to 0.96) (Supplementary Fig. 8).
Sensitivity analysis
The mitochondria-related indicators differed between compensated and decompensated cirrhotic patients. Therefore, stratified analyses were performed. The advantage of the ability to predict BI was observed in patients with decompensated cirrhosis (Table 2C), while for compensated patients, the MMP of CD4−CD8− T cells showed better diagnostic efficiency (Table 2B).
The MMP of CD8+ T cells demonstrated the best performance in predicting BI and ACLF in the total cohort. To better understand which aspects of CAIDs were reflected by the MMP of CD8+ T cells, we performed correlation analyzes with markers of inflammation, immune exhaustion, liver disease, and complications (Supplementary Fig. 9). These results suggested that the MMP of CD8+ T cells was associated with age and the stage of liver disease and that the MMP of CD8+ T cells was significantly correlated with inflammation-related markers.
We then performed subgroup analysis based on the stage of cirrhosis. Although the HR varied among subgroups, the MMP of CD8+ T cells demonstrated strong predictive ability for BI in subgroups. The BI-free rate differed between compensated patients with high and low MMP of CD8+ T cells (hazard ratio 2.95, 95% CI 1.12–7.74; P = 0.01; Fig. 1B). In the absence of positive events, the hazard ratio could not be calculated (Supplementary Fig. 6B). BI was more common in decompensated cirrhotic patients with high MMP of CD8+ T cells (hazard ratio 3.35, 95% CI 2.15–4.97; P < 0.001; Fig. 1C). Decompensated patients with high MMP were more likely to develop ACLF(hazard ratio 2.73, 95% CI 1.30–5.71; P < 0.001; Fig. 2C). Compared with the traditional index, the results of the subgroup analysis suggested that the advantage of MMP of CD8+ T cells was mainly in decompensated patients (Supplementary Table 2C).
The results of stratification analysis according to ages were similar to the main findings (Supplementary Table 3).
Figure 2 showed the persistent association of the MMP of CD8+ T cells with the occurrence of BI and ACLF after adjustment for age and cirrhosis stage. We observed that the lower limit of the hazard ratio with the 95% CI was greater than 1 at the MMP of less than 37.3 in Figure 2A. There was a nonlinear relationship between MMP and ACLF risk (nonlinear P < 0.05, Fig. 2B).
Mitochondria damage and abnormally activated mitochondrial autophagy in CD8+ T cells
The majority of mitochondrial structures in CD8+ T cells from patients with high MMP were roughly normal (Fig. 3A). In patients with CD8+ T cells with low MMP, abnormal mitochondria were observed (mitochondrial cristae contracted along the outer membrane, and some mitochondrial cristae were unclear), and damaged mitochondria accumulated. The mitochondrial phagosomes stuck to lysosomes, with damaged mitochondria being engulfed by lysosomes, and the mitochondrial membrane protruded into vesicle-like structures (Fig. 3B). Many typical autophagy structures were observed under transmission electron microscopy (Fig. 3B).
Then we collected CD8+ T cells for mitochondrial autophagy-related protein detection, which revealed that the levels of autophagy-related protein 8a (LC3) and classical mitochondrial autophagy pathway proteins PINK1 and Parkin accumulated in patients with low-MMP T cells, while the ubiquitin-binding scaffold protein (p62) was significantly downregulated (Supplementary Fig. 10A). When we measured the intracellular ROS content in the CD3+ T cells, the results indicated that the intracellular ROS content in the T cells of patients with low MMP was significantly decreased (Supplementary Fig. 10B). These findings suggest that T cells with low MMP have abnormal mitochondrial autophagy activation.
Patients with low CD8+ T-cells MMP had high systemic inflammatory response
Supplementary Figure 9 showed that the MMP of CD8+ T cells was significantly correlated with inflammation-related markers. We further collected peripheral blood samples for cytokine assays to assess whether these markers reflected systemic inflammatory changes in CAIDs. TNF-α and IL-6 levels were also found to be moderately correlated with the MMP of CD8+ T cells (0.3<r<0.5, P < 0.05, Supplementary Fig. 11).
Discussion
Patients with cirrhosis have a unique spectrum of immune changes that characterize the course of end-stage liver disease.9 Immune deficiency is one of the key manifestations of CAID.8 The severity of CAID is related to the stage of cirrhosis.9 The progression of cirrhosis has a significant impact on the intensity of CAID,30,31 which is associated with the severity of liver insufficiency, bacterial translocation, and organ failure.32 Patients with compensated cirrhosis and stable decompensated cirrhosis exhibit an overactive immune system, but immune function, including the ability to fight infection, is not significantly impaired. However, in patients with ACLF, there is significant immune dysfunction, which increases the risk of infection and worsens prognosis.9,31 It was previously thought that the overall effector response of cirrhotic patients in the compensated and stable-decompensated stages was not significantly impaired. However, there is a marked difference in the incidence of BI between the two stages. While both compensated and stable decompensated patients exhibit similar low-grade systemic inflammatory phenotypes,9 they differ in T-cell mitochondrial function. In our study, we found that the MM of total T cells, CD4+ T cells, and CD8+ T cells increased, while MMP decreased significantly during decompensation.
T-cell depletion occurs in patients with ACLF.33,34 Throughout the life cycle of immune cells, energy and substrate requirements change dramatically, and metabolic pathways are activated or inhibited to adapt to cell development and activation.35 Mitochondrial changes occur in T cells with different phenotypes and at various stages of the T-cell life cycle.36,37 Therefore, we focused on T-cell mitochondrial function to clarify the relationship between T-cell mitochondrial function indicators and BI in patients with cirrhosis. We recorded the MM and MMP of patients’ total T cells, CD4+ T cells, CD8+ T cells, CD4+CD8+ T cells, and CD4+CD8+ T cells. Among many T-cell mitochondrial indicators, the MMP of CD8+ T cells showed the best performance in predicting BI and ACLF and surprisingly performed better than traditional indicators. The multivariable Cox regression model showed that the MMP of CD8+ T cells was an independent risk factor for BI.
Our study found that the MMP of CD8+ T cells was significantly and weakly correlated with age and liver disease stages. This is in line with previous literature, which suggests that CAIDs are related to liver disease stages and age, indicating that the MMP of CD8+ T cells might reflect the severity of CAIDs. Sensitivity analyses were performed to account for the possible confounding role of age and liver disease stage. The RCS curves, adjusted for age and liver disease stage, still showed that a decreased MMP of CD8+ T cells was associated with an increased risk of BI. Subgroup analysis revealed that the advantage of the MMP of CD8+ T cells, compared with traditional inflammation indices, was mainly observed in patients with decompensated cirrhosis and was not as pronounced in compensated cirrhotic patients.
CAIDs mainly include immune exhaustion and abnormal systemic inflammation.9,31 Alterations in T cells have been shown to be associated with the development of secondary infections in patients with immune paralysis and prolonged sepsis.38 Our study suggests that the MMP of CD8+ T cells is related to markers of inflammation, but the role of MMPs in CAIDs still requires further study.
Mitochondria are involved in the mechanism of autophagy, and damaged mitochondria caused by minor stress can be cleared through mitochondrial autophagy.39 Intense stress can lead to an increase in the number of osmotic transition pores in the mitochondrial membrane, which affects mitochondrial autophagy.40 Abnormally activated mitochondrial autophagy in cells is characterized by decreased membrane potential.41 Abnormal mitochondria accumulated and mitochondrial autophagy increased in peripheral blood T cells of patients with low MMP in CD8+ T cells, as observed under electron microscopy. Protein detection indicated that, compared with patients with high MMP in CD8+ T cells, peripheral circulating total T cells and CD8+ T cells in patients with low MMP CD8+ T cells showed activation of abnormal mitochondrial autophagy. The low MMP of T cells in cirrhotic patients could be partially explained by mitochondrial morphological abnormalities and abnormal autophagy activation. The underlying mechanism still needs to be further explored.
We identified a novel biomarker, the MMP of CD8+ T cells, for the early identification of BI and ACLF in cirrhotic patients within 90 days of hospitalization. The results of the subgroup analysis suggested that the advantage of the MMP of CD8+ T cells in predicting BI and ACLF was primarily observed in decompensated patients. The MMP of CD8+ T cells was significantly and weakly correlated with inflammation-related markers. Our study suggests that mitochondrial abnormalities in T cells in peripheral blood might be actively involved in the pathogenesis of CAID, which will be the focus of our upcoming research. Further studies are also needed to determine whether prophylactic use of antibiotics or exogenous administration of immune molecules can improve prognosis in cirrhotic patients with low-MMP CD8+ T cells.
The limitations of this study are as follows: (1) Due to the nature of this single-center observational study, selection bias was inevitable; (2) Most patients in the compensated stage were hospitalized for pathological examinations or to assess complications, while most patients in the decompensated stage were hospitalized for ascites, occult or microhepatic hepatic encephalopathy, and jaundice. This prevents us from extending the conclusions of the study to post-EGVB patients (only two were hospitalized for EGVB); (3) Our study does not determine whether the MMP of CD8+ T cells can predict ACLF in cirrhotic patients with BI, as patients who suffered BI within one week and received antibiotics were excluded at the beginning of the study; (4) Although the MMP of CD8+ T cells demonstrated a reasonable ability to predict transplant-free survival, continued follow-up is needed because there were too few outcome events to draw definitive conclusions; (5) Due to ethical requirements (blood samples collected for TME exceeding 40 mL), our results in the TME were not detailed enough to extend to clinically unstable patients; (6) The cohort was not established with the concept of recompensation in mind; and (7) Our findings cannot be extended to patients with liver cancer or other organ failures.