Preoperative MRI-derived fat fraction is linked to survival after liver transplantation for hepatocellular carcinoma

Abstract

Background and aims

Liver transplantation (LT) is a curative option for hepatocellular carcinoma (HCC). However, HCC recurrence and the limited accuracy of current morphological criteria remain unresolved issues. This study investigated the association between preoperative magnetic resonance imaging (MRI)–derived fat fraction (FF) and survival in patients with HCC undergoing LT.

Methods

We retrospectively reviewed 42 patients with HCC who underwent orthotopic LT. Preoperative FF was obtained from dual-echo T1-weighted imaging, and patients were categorized into low- (n = 24) or high-FF groups (n = 18) using a 6.074% threshold. Clinical data, perioperative features, and outcomes were analyzed. Disease-free survival (DFS) and disease-specific survival (DSS) were assessed using Cox regression and Kaplan–Meier analyses.

Results

Baseline characteristics were comparable between the two groups, except for higher cholesterol level in the low-FF group. Elevated alpha-fetoprotein level (≥40 ng/mL) was independently associated with shorter DFS (hazard ratio [HR] = 3.649; 95% confidence interval [CI]: 1.002–13.289; p < 0.05). For DSS, FF emerged as an independent protective factor (HR = 0.848; 95% CI: 0.746–0.965; p = 0.012). Glucose level (HR = 0.852; 95% CI: 1.118–3.066; p = 0.017) and increased gamma-glutamyl transferase level (HR = 1.005; 95% CI: 1.001–1.008; p = 0.006) were linked to poorer DSS. Kaplan–Meier analysis demonstrated longer DSS in the high-FF group, with no differences in DFS.

Conclusion

Preoperative MRI-derived FF is independently associated with DSS after LT in patients with HCC. FF assessment may complement the Milan criteria and improve candidate selection, thereby supporting clinical decision-making.

Highlights

• •Preoperative MRI-derived FF predicts disease-specific survival after LT.
• •Higher tumor FF is independently associated with improved DSS after LT.
• •MRI-derived FF provides prognostic information beyond tumor morphology.
• •FF may complement the Milan criteria for risk stratification before LT.

Abbreviations

AFP

alpha-fetoprotein

ALB

albumin

ALP

alkaline phosphatase

ALT

alanine aminotransferase

AST

aspartate aminotransferase

BMI

body mass index

DFS

disease-free survival

FF

fat fraction

GGT

gamma-glutamyl transferase

HB

hemoglobin

HBV

hepatitis B virus

HCC

hepatocellular carcinoma

IP

in-phase

KM

Kaplan–Meier

LT

liver transplantation

MRI

magnetic resonance imaging

MVI

microvascular invasion

TNM

tumor node metastasis

OP

opposed-phase

DSS

disease-specific survival

PLT

platelet count

PT

prothrombin time

ROI

region of interest

TB

total bilirubin

TC

total cholesterol

TG

triglyceride

WBC

white blood cell count

1. Introduction

Hepatocellular carcinoma (HCC) is the fifth most common malignancy and the third leading cause of cancer-related mortality, with its incidence increasing worldwide.¹ Current treatment options for HCC include surgical resection, local ablation, and liver transplantation (LT).² LT provides unique advantages for patients with liver failure, as it simultaneously removes the tumor and replaces the diseased liver, thereby markedly improving long-term survival.^3,4 Nevertheless, postoperative recurrence remains a major clinical challenge, occurring in ∼10%–20% of cases.⁵ Consequently, accurate preoperative prognostic assessment is essential to optimize transplant candidate selection.

Since their introduction in 1996, the Milan criteria, based primarily on tumor morphology such as size, number, and vascular invasion, are considered the gold standard for determining transplant eligibility.^6,7 Despite being effective in reducing recurrence risk, these morphological parameters do not fully capture tumor biology and thus have limited prognostic accuracy.^8,9 To address this gap, increasing efforts have been directed toward developing novel predictive models and identifying robust biomarkers to improve risk stratification.

Various strategies, including liquid biopsy and advanced imaging techniques, have been investigated to improve prognostic assessment after LT 10, 11, 12. Early identification of patients at high risk of post-transplant recurrence can provide valuable information for clinical decision-making and facilitate precision oncology. Among these approaches, magnetic resonance imaging (MRI) has been increasingly applied in HCC owing to its superior soft tissue resolution and multiparametric capabilities.¹³ Notably, it offers a noninvasive and time-efficient evaluation, highlighting its potential for prognostic prediction in LT candidates with HCC. Previous studies have reported that the fat fraction (FF) derived from dual-echo–based sequences allows quantitative assessment of intratumoral lipid content.^14,15 In addition, accumulating evidence suggests that tumor FF is strongly linked to the biological characteristics of HCC, such as histological subtype, Ki67 expression, and degree of differentiation.^14,16,17 Moreover, intratumoral lipid content may influence recurrence risk, thereby affecting post-transplant survival outcomes.

Therefore, this study aimed to investigate the association between preoperative MRI-derived FF and survival in patients with HCC undergoing LT and evaluate its potential as a noninvasive imaging biomarker for optimizing patient selection and guiding personalized treatment strategies. In addition, we aimed to explore differences between high- and low-FF populations at the transcriptome level.

2. Material and methods

2.1. Patients

We conducted a retrospective study involving 500 patients with HCC who underwent orthotopic LT at Nanjing Drum Tower Hospital between January 2018 and September 2020 and at the Jinling Hospital, Affiliated Hospital of Medical School, Nanjing University, between October 2020 and January 2023. Exclusion criteria were as follows: (1) receipt of pre- or postoperative antitumor therapies, including transarterial chemoembolization, ablation, immunotherapy, or targeted therapy; (2) absence of MRI data within 2 weeks before surgery; (3) loss to follow-up after transplantation; (4) incomplete clinical or pathological data; or (5) hyperacute or acute graft rejection post-transplantation. After applying these criteria, 42 patients were included in the final analysis (Fig. 1). The study was approved by the Medical Ethics Committee of Drum Tower Hospital, Nanjing University School of Medicine (2023-LCYJ-PY-32), and written informed consent was obtained from all patients.

Fig. 1.

Patient flowchart.

2.2. Clinical data

Preoperative clinical and laboratory variables collected in this study included age, sex, body mass index (BMI), history of hypertension or diabetes, hepatitis B virus (HBV) status, and serum alpha-fetoprotein (AFP, ng/mL) level. Hematological and biochemical parameters included white blood cell count (10⁹/L), platelet count (10⁹/L), aspartate aminotransferase level (U/L), alanine aminotransferase level (U/L), alkaline phosphatase level (U/L), gamma-glutamyl transferase (GGT, U/L) level, albumin level (g/L), total cholesterol level (mmol/L), triglyceride level (mmol/L), fasting blood glucose level (mmol/L), and prothrombin time (s). Intraoperative variables included operation time (min), estimated blood loss (mL), volume of blood transfusion (mL), and anhepatic phase duration (min). The Milan criteria are based on the diameter and number of tumors observed on preoperative MRI.

2.3. MRI data

MRI was performed within 2 weeks before LT using a 3.0-T scanner (uMR770, United Imaging Healthcare, Shanghai, China) according to a standardized liver MRI protocol. The protocol included axial T2-weighted fat-suppressed imaging, dual-echo T1-weighted in-phase (IP) and opposed-phase (OP) imaging (slice thickness, 3 mm), diffusion-weighted imaging (b values of 50 and 800 s/mm²), and dynamic contrast-enhanced imaging after intravenous administration of Gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Primovist, Bayer HealthCare, Berlin, Germany). For the dual-echo sequence, the repetition time was 3.8 ms, and echo times were 1.38 (OP) and 2.30 (IP) ms. Detailed MRI acquisition parameters are provided in Supplementary Table S1. Regions of interest (ROIs) were manually delineated on the largest cross-sectional area of the tumor on IP and OP images. Contrast-enhanced MRI images in arterial, portal venous, delayed, and hepatobiliary phases were used as anatomical references during ROI placement to avoid hemorrhage as well as major vessels, necrotic regions, and imaging artifacts. FF was calculated according to a previously published method using the following formula: FF (%) = (1/2 − OP/2IP) × 100%, where OP and IP represent signal intensities on OP and IP images, respectively. FF was analyzed as a continuous variable in Cox regression analyses. Furthermore, patients were categorized into low- and high-FF groups using a predefined 6.074% cutoff for Kaplan–Meier survival analysis, according to our previous MRI study.¹⁴ Detailed MRI findings are provided in Supplementary method and Supplementary Table S1.

2.4. Pathology

All included cases were independently re-evaluated by two experienced pathologists specialized in hepatobiliary diseases. Both pathologists were blinded to clinical and imaging information during reassessment to ensure diagnostic consistency and minimize bias. Disagreements were resolved through joint discussion until consensus was reached.

2.5. Follow-up assessment

All patients underwent regular postoperative surveillance for HCC recurrence, including monthly serum AFP testing and imaging examinations every 3 months (contrast-enhanced computed tomography or MRI). Follow-up information was obtained through outpatient visits, inpatient records, and telephone interviews. Disease-free survival (DFS) was defined as the interval from LT to radiologically confirmed tumor recurrence, whereas disease-specific survival (DSS) was defined as the interval from surgery to death attributable to HCC. The median follow-up duration was 1181.5 (range, 112–2718) days. During follow-up, 11 patients experienced tumor recurrence and 11 died.

2.6. Statistical analysis

Continuous variables were presented as mean ± standard deviation for normally distributed data or as median (interquartile range, P25–P75) for non-normally distributed data. Between-group comparisons were performed using the independent samples t-test or Mann–Whitney U test, as appropriate. Categorical variables were expressed as counts and percentages and compared using the chi-square test or Fisher's exact test. Survival analyses were conducted using a univariate Cox proportional hazards model. To minimize model overfitting, variables with clinical relevance and p-values < 0.1 in the univariate analysis were entered into the multivariate Cox model. Variables with p-values < 0.05 in the multivariate analysis were considered independent predictors of DFS or DSS. Post-hoc power analyses for DSS and DFS were performed based on Freedman approximation for the log-rank test using the observed hazard ratios (HRs) and event numbers. To preserve quantitative information, FF was analyzed as a continuous variable in Cox regression analyses. Kaplan–Meier analysis and the log-rank test were used to evaluate the association between FF and DFS or DSS. All statistical analyses were performed using R version 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria), and a two-sided p-value < 0.05 was considered to indicate statistical significance.

3. Results

Table 1 summarizes baseline characteristics of the 42 included patients. The cohort comprised 24 patients in the low-FF group and 18 in the high-FF group. No significant differences were observed in age, sex, BMI, comorbidities, HBV status, tumor burden according to the Milan criteria, serum AFP levels, liver function parameters, or perioperative variables between the two groups.

Table 1.

Baseline data of enrolled patients.

Variable	Overall (n = 42)	Low FF (n = 24)	High FF (n = 18)	p
Age (years, mean ± SD)	53.67 ± 10.40	52.46 ± 11.66	55.28 ± 8.49	0.391
Sex (%)				0.949
Male	36 (85.7)	20 (83.3)	16 (88.9)
Female	6 (14.3)	4 (16.7)	2 (11.1)
BMI (kg/m2)	23.85 (22.75, 25.11)	23.07 (21.20, 25.31)	24.22 (23.52, 24.94)	0.269
Hypertension (%)				0.577
No	31 (73.8)	19 (79.2)	12 (66.7)
Yes	11 (26.2)	5 (20.8)	6 (33.3)
Diabetes (%)				0.212
No	33 (78.6)	21 (87.5)	12 (66.7)
Yes	9 (21.4)	3 (12.5)	6 (33.3)
HBV status (%)				0.955
No	8 (19.0)	4 (16.7)	4 (22.2)
Yes	34 (81.0)	20 (83.3)	14 (77.8)
AFP level (≥40 ng/mL) (%)				0.668
	27 (65.9)	14 (60.9)	13 (72.2)
≥40	14 (34.1)	9 (39.1)	5 (27.8)
WBC (×109/L)	4.20 (3.30, 5.55)	4.05 (3.25, 4.80)	4.40 (3.62, 5.85)	0.269
PLT (×109/L)	90.50 (57.25, 137.75)	84.00 (56.50, 140.50)	108.76 (70.00, 136.50)	0.620
HB (g/L)	132.04 (120.25, 146.25)	136.00 (121.75, 147.25)	128.04 (114.00, 139.50)	0.492
AST (U/L)	38.20 (22.30, 49.48)	38.70 (20.90, 48.45)	38.05 (26.50, 57.52)	0.351
ALT (U/L)	44.90 (26.40, 65.40)	34.20 (25.75, 59.70)	48.70 (36.90, 65.52)	0.189
ALP (U/L)	119.00 (72.70, 148.10)	119.28 (71.78, 145.02)	114.70 (74.90, 153.20)	0.903
GGT (U/L)	51.80 (29.50, 110.15)	51.80 (29.90, 117.00)	54.80 (29.83, 101.67)	0.632
TB (μmol/L)	24.30 (14.50, 34.70)	17.10 (14.05, 31.25)	27.95 (16.42, 40.85)	0.511
ALB (g/L)	38.10 (33.97, 41.43)	36.05 (34.17, 39.70)	38.85 (33.97, 41.79)	0.315
TC (mmol/L)	3.73 (2.85, 4.86)	4.20 (3.48, 5.07)	3.31 (2.70, 3.81)	0.064
TG (mmol/L)	0.91 (0.66, 1.14)	0.92 (0.84, 0.98)	0.90 (0.65, 1.22)	0.863
Glucose (mmol/L)	5.21 (4.36, 5.79)	4.93 (4.54, 5.66)	5.32 (4.34, 5.89)	0.725
PT (s)	13.20 (11.97, 14.38)	13.10 (12.10, 14.25)	13.20 (11.80, 15.60)	0.849
Milan criteria (%)				0.562
Within	22 (52.4)	14 (58.3)	8 (44.4)
Beyond	20 (47.6)	10 (41.7)	10 (55.6)
Operation time (min)	385.00 (302.75, 501.25)	372.50 (316.25, 482.50)	407.50 (288.75, 532.50)	0.675
Blood loss (mL)	1550.00 (1000.00, 2375.00)	1900.00 (975.00, 2925.00)	1500.00 (1000.00, 1675.00)	0.320
Blood transfusion (mL)	2000.00 (1325.00, 3200.00)	2000.00 (1575.00, 3200.00)	2075.00 (1206.25, 3100.00)	0.541
Anhepatic phase (min)	47.50 (35.50, 50.00)	45.50 (33.75, 50.00)	49.00 (40.50, 50.00)	0.460
Tumor number (%)				0.786
Single	33 (78.6)	18 (75.0)	15 (83.3)
Multiple	9 (21.4)	6 (25.0)	3 (16.7)
Tumor diameter (cm)	3.90 (2.43, 6.00)	3.00 (2.00, 5.03)	5.35 (3.38, 6.38)	0.059
Edmondson–Steiner grade (%)				>0.999
I–II	18 (42.9)	10 (41.7)	8 (44.4)
III	24 (57.1)	14 (58.3)	10 (55.6)
MVI (%)				>0.999
Yes	7 (16.7)	4 (16.7)	3 (16.7)
No	35 (83.3)	20 (83.3)	15 (83.3)
TNM stage				0.333
I		15 (62.5)	14 (77.8)
II–III		9 (37.5)	4 (22.2)

Abbreviations: FF, fat fraction; BMI, body mass index; HBV, hepatitis B virus; AFP, alpha-fetoprotein; WBC, white blood cell count; PLT, platelet count; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, gamma-glutamyltransferase; ALB, albumin; TC, total cholesterol; TG, triglyceride; PT, prothrombin time; MVI, microvascular invasion; TNM, tumor node metastasis; HB, hemoglobin; SD, standard deviation.

In univariate Cox regression analysis for DFS, serum AFP level (HR = 4.519; 95% confidence interval [CI]: 1.317–15.510; p = 0.017), tumors beyond the Milan criteria (HR = 3.588; 95% CI: 0.946–13.609; p = 0.060), and HBV positivity (HR = 0.289; 95% CI: 0.082–1.020; p = 0.054) were associated with DFS (Fig. 2). In multivariate analysis, elevated AFP level remained an independent predictor of shorter DFS with borderline significance (HR = 3.649; 95% CI: 1.002–13.289; p < 0.05; Fig. 2).

Fig. 2.

Univariate and multivariate Cox regression analyses of disease-free survival in patients with HCC after liver transplantation. Abbreviations: BMI, body mass index; HBV, hepatitis B virus; AFP, alpha-fetoprotein; WBC, white blood cell count; PLT, platelet count; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, gamma-glutamyltransferase; ALB, albumin; TC, total cholesterol; TG, triglyceride; PT, prothrombin time; FF, fat fraction; HR, hazard ratio; CI, confidence interval; TNM, tumor node metastasis.

Univariate Cox regression analysis identified FF as a significant protective factor for DSS (HR = 0.885; 95% CI: 0.799–0.981; p = 0.020). Other factors associated with DSS included glucose level (HR = 1.424; 95% CI: 0.946–2.143; p = 0.091) and GGT level (HR = 1.005; 95% CI: 1.001–1.008; p = 0.006; Fig. 3). Multivariate Cox regression analysis confirmed FF as an independent protective factor for DSS (HR = 0.848; 95% CI: 0.746–0.965; p = 0.012), whereas elevated glucose level (HR = 1.852; 95% CI: 1.118–3.066; p = 0.017) and GGT level (HR = 1.005; 95% CI: 1.001–1.008; p = 0.006) were identified as independent predictors of poor survival (Fig. 3). Kaplan–Meier survival curves showed that patients with high FF had significantly longer DSS than those with low FF (log-rank p < 0.05; Fig. 4A), with no significant difference in DFS observed between the two groups (Fig. 4B). No significant differences in DFS or DSS were observed between the two groups in either the within- or beyond-Milan subgroups (Supplementary Fig. S1). Representative cases with high and low intratumoral FF are shown in Fig. 5.

Fig. 3.

Univariate and multivariate Cox regression analyses of disease-specific survival in patients with HCC after liver transplantation. Abbreviations: BMI, body mass index; HBV, hepatitis B virus; AFP, alpha-fetoprotein; WBC, white blood cell count; PLT, platelet count; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, gamma-glutamyltransferase; ALB, albumin; TC, total cholesterol; TG, triglyceride; PT, prothrombin time; FF, fat fraction; HR, hazard ratio; CI, confidence interval; TNM, tumor node metastasis.

Fig. 4.

Prognostic analysis of patients with HCC after liver transplantation. Kaplan–Meier curves for disease-specific survival (A) and disease-free survival (B) demonstrating a significant difference (log-rank test) between the high-FF and low-FF groups.

HCC, hepatocellular carcinoma; FF, fat fraction.

Fig. 5.

Representative magnetic resonance imaging images of HCC patients with high and low intratumoral FF. Case 1: A 52-year-old male with a high intratumoral FF of 15.486% and an AFP level of 35.7 ng/mL. The patient underwent liver transplantation and remained recurrence-free for 1.5 years after transplantation. The patient was Child–Pugh class A. (A) out-of-phase image, (B) in-phase image, (C) water map, (D) fat map, (E) fat-suppressed T2-weighted image, (F) diffusion-weighted image, (G) arterial phase image, and (H) portal venous phase image. Case 2: A 63-year-old female with a low intratumoral fat fraction of 3.511% and an AFP level of 1.8 ng/mL. The patient underwent liver transplantation and developed tumor recurrence 8 months after transplantation. At the time of recurrence, the patient was Child–Pugh class A and died 1.5 years after transplantation. (I) out-of-phase image, (J) in-phase image, (K) water map, (L) fat map, (M) fat-suppressed T2-weighted image, (N) diffusion-weighted image, (O) arterial phase image, and (P) portal venous phase image. HCC, hepatocellular carcinoma; FF, fat fraction.

4. Discussion

To the best of our knowledge, this is the first study to demonstrate that preoperative MRI-derived FF is significantly associated with DSS, but not with DFS, in patients with HCC undergoing LT. In addition, other clinical and laboratory parameters, including AFP, glucose, and GGT levels, showed prognostic relevance. These findings collectively provide novel insights into risk stratification of transplant candidates and warrant comparison with prior studies, which could provide additional biological and prognostic information beyond traditional morphological criteria.

Mechanistically, intratumoral lipid content reflects, to some extent, the molecular subtype and biological behavior of HCC. Previous studies have shown that lipid-rich HCC (particularly steatohepatitis HCC) is often associated with lower Ki67 expression, better differentiation, and less microvascular invasion—factors closely associated with favorable long-term survival 14, 15, 16, 17. Therefore, patients with elevated FF in our cohort might possess a relatively indolent tumor biology, potentially contributing to prolonged survival after transplantation. This metabolic shift and its interaction with the liver microenvironment may reflect altered tumor cell differentiation.¹⁵ Although these underlying molecular and histopathological mechanisms were not directly evaluated in our study, the abovementioned literature provides a strong biological rationale for the independent prognostic value of FF in predicting DSS, potentially supplementing the Milan criteria.

Notably, this study found a significant correlation of FF with DSS, but not with DFS. This discrepancy may be due to the following reasons. First, the limited sample size resulted in insufficient statistical power, which may have underdetected true differences. Second, FF is more likely to reflect a patient's overall metabolic status and mild tumor biology. This may explain its stronger association with long-term survival than with the risk of tumor recurrence. Third, postoperative recurrence is influenced by multiple factors, including immune status, anti-rejection therapy, and HBV reactivation, which may mask the contribution of FF to DFS. Therefore, further validation of the heterogeneous role of FF in predicting DFS and DSS is warranted in larger, multicenter cohorts.

Beyond FF, our study identified several established prognostic factors, which are consistent with those described in previous reports. Elevated AFP level was associated with inferior DFS, in line with previous evidence that AFP is a robust biomarker for recurrence and survival after transplantation 18, 19, 20. Furthermore, elevated fasting blood glucose level can affect recurrence and long-term DSS in patients with HCC after transplantation.²¹ In addition, elevated GGT level independently predicted poor DSS, corroborating prior studies linking high GGT levels to aggressive tumor biology, microvascular invasion, and poor survival.^22,23 These findings suggest that combining traditional markers such as AFP and GGT with novel imaging biomarkers such as FF could provide a more comprehensive prognostic assessment beyond the Milan criteria, thereby improving risk stratification and guiding individualized patient management.

This study has several limitations. First, this was a single-center retrospective analysis with a small sample size and excluded patients who received postoperative antitumor therapies. Therefore, this study may be subject to selection bias. Given the limited number of outcome events, multivariate Cox models should be interpreted with caution because overfitting cannot be completely excluded. Post-hoc power analysis showed that the study had adequate statistical power for DSS analysis but limited power for DFS analysis, suggesting that the negative DFS findings should be interpreted cautiously, thus warranting validation in larger cohorts. Second, because multiple clinical, pathological, and imaging variables were evaluated simultaneously against dual endpoints (DFS and DSS), the lack of formal multiple comparison corrections (such as Bonferroni or false discovery rate) may increase the risk of type I errors (false positives); thus, the resulting p-values should be interpreted as exploratory. Third, FF measurement relies on manual ROI delineation. This method provides a simplified FF estimation and may be influenced by T2∗ effects, iron deposition, magnetic field inhomogeneity, and image noise. More advanced quantitative techniques, such as multi-echo Dixon or proton density FF, may provide more accurate fat quantification and should be used in future studies. Fourth, the lack of in-depth analysis of tumor molecular characteristics and metabolic pathways prevented direct validation of the association between FF and molecular subtypes or genes related to lipid metabolism. Last, several important pathological prognostic factors, including capsule invasion and viable tumor burden, were not comprehensively analyzed in the present study. Future multicenter studies with larger and more representative cohorts are warranted to validate our findings.

5. Conclusion

Our findings suggest that preoperative MRI-derived FF is a potential prognostic imaging biomarker for DSS after LT in patients with HCC. In this cohort, higher preoperative FF was associated with improved DSS. However, these findings should be interpreted with caution and require validation in larger, prospective, multicenter studies.

CRediT authorship contribution statement

Jialin Gao: Writing – original draft, Visualization, Formal analysis, Data curation. Donghua Zhang: Visualization, Validation, Formal analysis. Tao Jiang: Methodology, Formal analysis, Data curation. Decai Yu: Writing – review & editing, Visualization, Funding acquisition.

Ethics statement

The study was approved by the Medical Ethics Committee of Drum Tower Hospital, Nanjing University School of Medicine (2023-LCYJ-PY-32), and written informed consent was obtained from all patients. Jinling hospital formally recognised and accepted the lead center's ethical approval.

Organ donation

Not applicable.

Declaration of generative AI and AI-assisted technologies in the writing process

Not applicable.

Funding

This work was supported by grants from the National Natural Science Foundation of China [No. 82372834, 82173129, and 82373911]. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Informed consent

Written informed consent was obtained from all patients.

Data availability statement

Data will be made available by the corresponding author upon reasonable request.

Animal treatment

Not applicable.

Declaration of competing interest

The author Decai Yu is an Editorial Board Member for this journal and was not involved in the editorial review or the decision to publish this article.

Appendix A. Supplementary data

The following is the Supplementary data to this article.