Category: 3. Business

This study was approved by the University of NSW (UNSW) Sydney Human Research and Ethics Committee: number HC220624.

Study design

Intervention components and adaptation

The Raregivers Wellness Retreat programme was initially designed by Raregivers Inc to support family members caring for a loved one impacted by a rare disease. It included three virtual sessions per week delivered over 6 weeks, focusing on guided exercises, sensory self-care experiences, and the Raregivers Emotional Journey Map. The program also incorporated Sustainable Carer Workshops, providing practical strategies for self-care and resilience. Participants engaged in group-based sessions via Zoom™, ensuring flexibility and accessibility. Each session was designed to last approximately 60 min, with additional resources and exercises provided for participants to complete at their own pace throughout the week. The program was free to participants, and each carer received a wellness kit containing educational and self-care materials to enhance their experience.

Adaptation of the program for the Australian context was led by consumer lead (KP) and Raregivers course coordinator (ML), who have lived experience caring for children with complex rare epilepsy syndromes. They co-designed the weekly program to ensure its relevance and acceptability for Australian participants. While the core structure of the program remained consistent with the original design, the adaptation integrated culturally and contextually relevant topics. The co-design process ensured that the program addressed the unique challenges faced by carers in Australia, enhancing its impact and resonance with participants. This iterative approach allowed sessions to evolve in response to participant needs, ensuring the program was both dynamic and participant-centred.

Participants

Carers were eligible to participate in the study if they had attended the retreat and met the following criteria: (i) were the primary carer of a child with a complex, rare epilepsy; (ii) were aged 18 years or older; (iii) were able to communicate in English; and (iv) had access to the internet and a computer. Carers were screened for psychological distress via the online expression of interest form, which incorporated a screening questionnaire with a validated Distress Thermometer (see Additional file 1) [14]. Because this was a pilot study, the Wellness Retreat was scoped to establish safety, feasibility, and acceptability. Individuals with high distress scores (≥ 9 on the Distress Thermometer) were therefore not included, as the program was designed as a strengths-based, preventative wellbeing initiative and was not equipped to provide acute psychological support. Eligible participants provided informed consent and were enrolled. Those screening with high distress were individually contacted by the research team, provided information on appropriate mental health services, and offered support to connect with these services if required.

Consent to participate in evaluation of the retreat was given electronically, and implied consent was received from participants who completed and submitted the baseline and follow-up questionnaires. Participants also gave verbal consent prior to starting the interviews.

Recruitment

Participants were recruited via the mailing lists and social media platforms (Twitter, Facebook, LinkedIn) of relevant not-for-profit and community organisations, including Genetic Epilepsy Team Australia (GETA), Epilepsy Foundation Australia, Rare Voices Australia, Genetic Alliance Australia, and Syndromes Without A Name Australia. These organisations shared the study advertisement, which included a link to the expression of interest form. Eligible carers received an email invitation to participate, which included a link to the study information sheet and consent form. A maximum of two follow-up attempts (email or telephone call, using contact details provided in the expression of interest form) were made to non-responders.

Evaluation design

The acceptability, feasibility and impact of the Raregivers Wellness Retreat on carers’ self-reported wellbeing and social inclusion was evaluated using a mixed-methods pilot study. Participants completed a pre-intervention (upon enrolment, before their first session; see Additional file 2) and post-intervention (1 week after the final session; see Additional file 3) questionnaire, administered online via the Research Electronic Data Capture (REDCap) platform [15, 16]. Questionnaires were anticipated to take approximately 30 min to complete.

Following the program, participants were invited to one-on-one semi-structured interviews via Zoom™. These interviews were audio recorded, de-identified, and transcribed for analysis (see Additional file 4 for the interview guide).

Measures

The primary outcomes of acceptability and feasibility were assessed post-intervention using 15 purpose-designed items (see Additional file 3).

Secondary outcomes related to participant wellbeing, quality of life, social inclusion and the extent to which participants live consistently with their values (valuing) were assessed using validated scales at pre- and post-intervention (Table 1 and Additional files 2 and 3).

Wellbeing

The 15-item PERMA Profiler questionnaire was utilised to assess wellbeing across five domains [17]. The full PERMA Profiler contains 23 items, however a shortened version with 15 items was used for this study.

Social inclusion

The Social Inclusion Scale (SIS) assesses individuals’ self-perceived levels of social inclusion [18,19,20]. The overall SIS score was computed by summing all responses (which range from 16 to 64). This scale formulates three elements negatively and scores them in reverse order when calculating the overall score.

Values

The Valuing Questionnaire was used to assess the extent to which participants lived consistently with their values [21]. The Progress subscale evaluates personal adherence to values, and the Obstruction subscale measures the extent of barriers hindering the pursuit of a valued life. The ratings for each subscale were aggregated.

Carer quality of life

The ASCOT-SCT4 [22] was used to assess carer experience and social care-related quality of life across 7 domains: (1) Occupation; (2) Control over daily life; (3) Self-care; (4) Personal safety; (5) Social participation; (6) Space and time to be yourself; (7) Feeling supported and encouraged.

Post-intervention interviews

At the conclusion of the program, participants were invited to participate in a one-on-one semi-structured interview that covered their overall impressions of the program, what they liked/disliked, challenges, feasibility and acceptability of the program as well as their wellbeing, self-efficacy and social connectedness. The interview guide (see Additional file 4) was developed by the multidisciplinary team which included an epilepsy nurse, genetic counsellor, psychologist, clinical geneticist, consumer engagement expert and implementation science experts.

Quantitative analyses

Data was subjected to descriptive analysis by a statistician independent to the implementation team. Various descriptive statistics were computed, including count, mean, standard deviation, proportion, median, and range. We conducted a comparative analysis of various indicators at two time points, baseline and post-intervention, to examine the changes. Inferential testing for effect were not undertaken due to the small sample size.

For the PERMA profiler [17], the domain specific change was calculated by subtracting participants’ baseline scores from their follow-up scores for each domain. A positive change score indicates improved wellbeing over time, while a negative change score indicates reduced wellbeing over time. We report this data as frequencies of change.

The SIS domain-specific change was computed by subtracting baseline scores from follow-up scores for each domain. Positive change scores suggest greater and stronger sense of social inclusion over time, while negative change scores indicate weaker and lower inclusion.

The domain-specific change for Valuing was calculated by subtracting baseline scores from follow-up scores in each domain. A positive change score in the Progress domain indicates a higher degree of valued living over time, whereas a higher score in Obstruction reflects a lower level of valued living.

For each domain of the ASCOT-SCT4, we compared the pre- and post-intervention scores for each participant. A positive change was defined as one or more category shifts towards an ‘ideal state’, and a negative shift as one or more category shifts towards ‘high level needs’ from baseline to follow-up.

Qualitative analysis

The responses to open ended questions and interviews were analysed using a thematic analysis approach to explore the common experiences of participating in the program and highlight the impact of the program, as well as any barriers and facilitators to implementation. All transcripts were coded and analysed in NVivo© Version 12. The analysis followed five key stages outlined by Braun et al. (2019) [23]: data familiarisation, generation of codes, searching for themes, grouping and reviewing themes, and defining and naming themes. The transcripts were first independently read and coded by two researchers (KP, JM). These initial codes were then reviewed and updated following a discussion with the research team (KP, JM, SB). After the initial coding process, themes were generated and discussed with the research team (KP, JM, EEP), including the relationships between themes, and revision and collapsing of overlapping themes.

Mixed methods analysis

The quantitative and qualitative data were integrated narratively at the interpretation and reporting level using a weaving approach [24]. The quantitative and qualitative findings are presented together on a theme-by-theme basis according to the outcomes.

Daniel Leal-Olivas/PA Wire

‘Solar rooftop initiative’

UK government borrowing in September hit the highest level for the month in five years, official figures show, highlighting the challenges facing the chancellor ahead of next month’s Budget.

Borrowing – the difference between public spending and tax income – was £20.2bn in September, up £1.6bn from the same month last year, the Office for National Statistics (ONS) said.

A rise in debt interest payments offset the increased amount the government had raised through tax and national insurance, the ONS said.

Chancellor Rachel Reeves is widely expected to raise taxes in November’s Budget in order to meet her self-imposed rules for government finances.

September’s figure was slightly less than analysts’ expectations of £20.8bn, but was just above the £20.1bn that had been projected by the government’s official forecaster, the Office for Budget Responsibility (OBR), in March.

Borrowing over the first six months of the financial year has now reached £99.8bn, which the ONS said was the second-highest total for that period since monthly records began in 1993, after that of 2020.

The OBR is set to update its forecasts next month – laying out how much of a shortfall the government will need to make up if it is to meet its own tax and spending rules by the end of the current parliament.

Speaking to the BBC’s Today programme, Capital Economics chief economist Paul Dales said the chancellor would “love tax receipts to be higher” but that it would depend on faster growth in the economy.

Capital Economics is projecting that the government will have to raise £27bn in the Budget, with “higher taxes on households having to do the heavy lifting”.

Nick Ridpath, research economist at the Institute for Fiscal Studies, also pointed out the challenges facing the chancellor, citing “high debt interest spending, tiny headroom and a looming productivity downgrade” as factors that could make things “even trickier”.

While the focus remains on what measures the chancellor might take in the Budget, on the financial markets UK government borrowing costs have been easing over the past couple of months.

In August, long-term borrowing costs reached their highest level since 1998, when the interest rate on 30-year government bonds, known as the yield, jumped to 5.7%. However, this has now fallen to below 5.3%.

The yield on 10-year UK government bonds – usually taken as the benchmark for the cost of borrowing – is now at about 4.5%, down from a 4.8% in August.

Ahead of the Budget, the government has announced a number of measures it is taking to try to boost economic growth.

At the Regional Investment Summit in Birmingham on Tuesday, Reeves will announce reforms to scrap paperwork and red tape, which she says will save companies almost £6bn a year.

The ONS figures showed that although tax income in September was higher than last year, in part due to the increase in employers’ national insurance contributions, spending also increased.

This was partly due to pay rises and inflation increasing the government’s day-to-day running costs, as well as inflation-linked increases to state benefits.

The government also had to pay £9.7bn in debt interest, which was up by £3.8bn from the same month last year.

Public sector debt in the UK is now estimated to be at 95.3% of gross domestic product (GDP) and remains at levels not seen since the early 1960s.

Responding to the figures, Chief Secretary to the Treasury James Murray said the government would “never play fast and loose with the public finances”.

He reiterated the government’s aim of bringing down borrowing, to be rid of “costly debt interest, instead putting that money into our NHS, schools and police”.

But shadow chancellor Mel Stride said that borrowing was “soaring under this Labour government”.

“Rachel Reeves has lost control of the public finances and the next generation are being saddled with Labour’s debts,” he said.

Liberal Democrat Treasury spokeswoman Daisy Cooper said “alarm bells should be ringing for the government ahead of the Budget”.

She said the Conservatives had left the economy in “a terrible state” but that “this government has made mistake after mistake, failing to get our economy growing again”.

The new figures from the ONS include a correction to earlier data, when an error had been made in how VAT receipts had been added.

Introduction

Hypertension, specifically essential hypertension (also known as primary or idiopathic hypertension), is a major contributing risk factor for cardiovascular diseases (CVDs) and other diseases with adverse clinical outcomes, making it the largest single contributor to morbidity and mortality worldwide.¹ Pre-hypertension, defined as the intermediate stage between normotension and hypertension, is also significantly correlated with a heightened risk of various cardiovascular outcomes including total CVD, coronary heart disease, myocardial infarction, and stroke.² Elevated blood pressure and its consequent cardiovascular and cerebrovascular diseases have become the predominant burden of disease and leading cause of mortality globally and in most regions of the world,³ affecting approximately 1.39 billion individuals and resulting in over 10.8 million deaths annually.⁴ Owing to the efforts of governments and health organizations and the widespread use of anti-hypertensive medications, global mean blood pressure has remained constant or decreased slightly over the past five decades. However, the prevalence of hypertension has still increased, particularly in low- and middle-income countries.⁵ Consequently, moving the prevention gateway forward and implementing targeted preventive measures at the pre-hypertension stage may offer a breakthrough in reducing the incidence of hypertension.

Obesity is proven to be an established risk factor for hypertension and CVDs,^6–8 and the close association of obesity with blood pressure has long been recognized in diverse populations.^9–11 As conventional anthropometric indicators of obesity, body mass index (BMI) is widely used in research on predictive factors of pre-hypertension and hypertension.^12,13 Waist circumference (WC) and waist-height ratio (WHtR) also show good potential to predict pre-hypertension and hypertension.¹⁴ BMI reflects the overall body fat distribution and cannot accurately distinguish between fat and muscle proportions; WC and WHtR indicate abdominal obesity accurately and cannot distinguish between subcutaneous fat and visceral fat. In addition, other conventional anthropometric indicators such as basal metabolic rate (BMR), fat mass (FM), fat-free mass (FFM), fat mass index (FMI), and fat-free mass index (FFMI) are also considered effective predictors of hypertension.^15–17 However, the effects of these conventional anthropometric indicators on pre-hypertension are still limited. Furthermore, only a few studies reported the relationship between body fat percentage (BFP), visceral fat index (VFI) and pre-hypertension,^18,19 but the conclusions among them seemed to be controversial.

Recently, a number of new anthropometric indicators have garnered increasing attention due to their higher value in predicting disease risk. Waist circumference index (WCI), weight-adjusted-waist index (WWI), body surface area (BSA), conicity index (CI), body roundness index (BRI), and a body shape index (ABSI) are parameters based on specific combinations and calculations of physical examination indicators such as height, weight, or WC, while atherogenic index of plasma (AIP) and triglyceride-glucose index (TyG) are derived from the arithmetic operation of biochemical indicators such as triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), or fasting plasma glucose (FBG). Lipid accumulation product (LAP), visceral adiposity index (VAI), and cardiometabolic index (CMI) combine physical examination indicators such as height or WC with biochemical indicators such as TG and HDL-C. Some studies showed these unconventional anthropometric indicators are associated with hypertension, pre-hypertension, or blood pressure.^20–22 Differing from most studies that consider only one single indicator, this study is one of the first studies to compare a broad panel (24 indicators) simultaneously in a large, population-based Chinese survey. It would be of great clinical and practical importance to further explore the best anthropometric indicators for identifying pre-hypertension or hypertension.

Therefore, this study aimed to identify and compare the predictive value of 24 conventional and unconventional anthropometric indicators for pre-hypertension and hypertension among Chinese adults.

Methods

Study Design and Population

In this study, we used data from the Fujian Province part of the Chinese Residents Cardiovascular Disease and Risk Factors Surveillance Project, 2020. The project was carried out on a stratified multi-stage random sampling method, taking into account the economic development level of urban and rural areas, and selected 262 nationally and provincially representative districts and counties in 31 provinces, autonomous regions, and municipalities (excluding Hong Kong, Macao, and Taiwan) on the basis of their administrative divisions as monitoring sites. At each monitoring point, 1200 permanent residents aged 18 and above were selected according to sex and age group. A total of about 300,000 people were eventually surveyed on the status and distribution characteristics of important CVDs and risk factors such as obesity, hypertension, diabetes mellitus, dyslipidemia, coronary heart disease and stroke, etc.

As a result, three cities in urban areas and five counties in rural areas were selected. A total of 9790 participants living in Fujian Province for more than six months and aged 18 years or older were randomly selected to participate in this survey from August 2020 to April 2021. The exclusion criteria were set as follows: (1) participants with missing systolic blood pressure (SBP)/diastolic blood pressure (DBP) data (n = 59); (2) participants with missing data on total cholesterol (TC), TG, HDL-C, low-density lipoprotein cholesterol (LDL-C), FBG, or uric acid (UA) (n = 943); (3) participants with negative and clearly erroneous values of FFM (n = 1). After data filtration, a total of 8787 participants were ultimately selected for subsequent analyses. This study followed STROBE reporting guidelines (see Supplementary Table 1).

Blood Pressure Measurement and Definition of Pre-Hypertension or Hypertension

BP was measured using the same brand and model of electronic sphygmomanometer (Omron electronic sphygmomanometer HBP-1120U), with an accuracy of ±1 mm Hg. Participants were uniformly measured on the right upper arm, and three measurements were recorded, with no more than a 10 mm Hg difference in systolic or diastolic readings between any two of the three measurements (1 mm Hg = 0.133 k Pa). SBP or DBP was defined as the average of the three SBP or DBP readings.

The classification of normotension, pre-hypertension and hypertension was based on the criteria from JNC-8.²³ Hypertension was defined as an average SBP ≥140 mm Hg, and/or average DBP ≥90 mm Hg, and/or previously diagnosed with hypertension, and/or currently taking anti-hypertensive drugs; pre-hypertension was defined as an average SBP in the range of 120 to 139 mm Hg and/or an average DBP in the range of 80 to 89 mm Hg, without being on anti-hypertensive drugs; normotension was defined as an average SBP <120 mm Hg and an average DBP <80 mm Hg, without being on anti-hypertensive drugs.

Anthropometric Measurements

Anthropometric measurements were performed by trained staff according to standard procedures. Height was measured using a stadiometer of the same model with a length of 2.0 m and a minimum scale of 1 mm (Suhong BT-24). WC was measured at the superior border of the iliac crests using a waist circumference ruler of the same brand and model with a length of 1.5 m, a width of 1 cm, and a minimum scale of 1 mm. Weight, FM, FFM, BFP, VFI, and BMR were measured using the same brand and model of weight and body fat measuring device (InBody H20B body composition analyzer), with the scale function measuring accurately to 0.1 kg and a maximum weighing capacity of 150 kg. The above parameters must be measured on an empty stomach in the early morning, and participants should be dressed in light clothes. BMI, height-adjusted weight (HtaW), WHtR, FMI, FFMI, WCI, WWI, BSA, AIP, LAP, VAI, TyG, CI, BRI, ABSI, and CMI were calculated according to previous published formulae as followed (HtaW was calculated based on the coefficients of the sex-specific linear regression models of weight on height):

²⁴

²⁵

²⁶

²⁷

²⁸

²⁹

³⁰

³¹

³²

³³

³⁴

³⁵

³⁶

³⁷

³⁸

Covariates

About 21 covariates were identified based on literature. Demographic characteristics, health-related behaviors, and family history were collected by a standardized questionnaire form through face-to-face interviews with trained staff and physical measurements. Demographic characteristics included age, sex, current residence, educational level, annual household income per capita, occupation, medical insurance, household size, marital status. Educational level was categorized into five levels: illiterate, primary school, junior high school, high school/technical secondary school, and junior college/undergraduate or above. Occupation was divided into three categories: unemployed, physical labor, and mental labor. Health-related behaviors included smoking, drinking, physical activity, regular sleep. Smoking was divided into three groups: never (never smoked in a lifetime), former (smoked in the past but not in the past 30 days), and current (smoked in the past 30 days). Drinking was also classified into three levels: never (never drank alcohol in a lifetime), former (used to drink alcohol but do not drink it anymore), and current (drank in the previous 30 days). Physical activity was defined as those who engaged in moderate-intensity or above physical activity for at least 5 days a week and at least 30 minutes per day during the past month. Regular sleep referred to those who slept between 7 and 9 hours per night on average in the past month. Family history involved hypertension and CVDs, and family history of CVDs was defined as one of the parents had coronary heart disease or stroke. Blood biochemical indicators included FBG, TC, TG, HDL-C, LDL-C, and UA, all tested using the Beckman AU680 instruments. The glucose oxidase method was used for FBG, the enzymatic method was used for TC and TG, the direct method was used for HDL-C and LDL-C, and the uricase method was used for serum UA.

Statistical Analysis

Continuous variables were presented as mean ± standard deviation or median (IQR) as appropriate and compared using the ANOVA, Welch ANOVA test or Kruskal–Wallis rank sum test, which depended on whether the quantitative data were consistent with the normal distribution. Categorical variables were expressed as percentages and analyzed by the Chi-square test or Fisher’s exact test as appropriate. Univariate logistic regression was used to judge the relationship between anthropometric indicators and pre-hypertension or hypertension, as well as a correlation matrix was calculated and a heat map was plotted to assess the multicollinearity of independent variables. Least absolute shrinkage and selection operator (LASSO) regression was employed to select independent variables, and multivariate logistic regression was further used to analyze the association between screened anthropometric indicators and pre-hypertension or hypertension, adjusting for confounders. Receiver-operating characteristic (ROC) curve was used to evaluate the discrimination ability of each screened anthropometric indicators for pre-hypertension or hypertension. The potential dose-response relationships between anthropometric indicators and the risk of pre-hypertension or hypertension was assessed by restricted cubic spline (RCS) models. A two-sided P-value <0.05 was considered significant. All statistical analyses were conducted using R 4.3.3 and DecisionLinnc. 1.0 softwares.

Results

Characteristics of the Study Population

Figure 1 showed a detailed flow chart of participants screening process. As shown in Table 1, among the 8787 participants, the prevalence of pre-hypertension and hypertension were 34.92% (3068/8787) and 35.84% (3149/8787), respectively. Compared with the normotension group, participants with pre-hypertension or hypertension tended to be older. The differences in sex, current residence, educational level, annual household income per capita, occupation, medical insurance, household size, marital status, smoking, drinking, regular sleep, family history of hypertension or CVDs within the blood pressure subgroups were statistically significant (all P < 0.05). Besides, the pre-hypertension and hypertension groups displayed notably higher FBG, TC, TG, LDL-C, UA and lower HDL-C in comparison to the normotension group (all P < 0.05).

Table 1 Characteristics of the Study Population

Figure 1 A detailed flow chart of participants screening process.

Distributions of Conventional and Unconventional Anthropometric Indicators

Table 2 showed the distributions of 24 conventional and unconventional anthropometric indicators in the normotension, pre-hypertension, and hypertension groups. There were statistically significant differences in height, weight, WC, FM, FFM, BFP, VFI, BMR, BMI, HtaW, WHtR, FMI, FFMI, WCI, WWI, BSA, AIP, LAP, VAI, TyG, CI, BRI, ABSI, and CMI among the blood pressure subgroups (all P < 0.05). Compared to the normotension group, the pre-hypertension group exhibited significantly higher height, weight, WC, FM, FFM, VFI, BMR, BMI, HtaW, WHtR, FMI, FFMI, WCI, WWI, BSA, AIP, LAP, VAI, TyG, CI, BRI, ABSI, and CMI; while the hypertension group showed significantly higher weight, WC, FM, FFM, BFP, VFI, BMR, BMI, HtaW, WHtR, FMI, FFMI, WCI, WWI, BSA, AIP, LAP, VAI, TyG, CI, BRI, ABSI, and CMI, but lower height (all P < 0.05).

Table 2 Distributions of Conventional and Unconventional Anthropometric Indicators in Different Blood Pressure Subgroups

Selection of Anthropometric Indicators by LASSO Regression

The results of univariate logistic regression analysis showed that 23 conventional and unconventional anthropometric indicators were associated with pre-hypertension (except for BFP), and all 24 conventional and unconventional anthropometric indicators were associated with hypertension (see Supplementary Table 2). However, the correlation matrix and heat map revealed multicollinearity among the independent variables (see Supplementary Table 3 and Supplementary Figure 1). Thus, we used LASSO regression to identify key anthropometric indicators associated with pre-hypertension or hypertension, respectively. As presented in Figure 2, the LASSO regression model identified 15 conventional and unconventional anthropometric indicators for pre-hypertension (including weight, FFM, VFI, BMR, BMI, HtaW, FMI, FFMI, WWI, AIP, LAP, VAI, TyG, BRI, and CMI), and 18 conventional and unconventional anthropometric indicators for hypertension (including height, WC, FM, BFP, VFI, BMR, BMI, HtaW, WHtR, FMI, FFMI, WWI, AIP, LAP, VAI, TyG, BRI, and CMI).

Figure 2 Selection of anthropometric indicators by LASSO Regression. Notes: (A) LASSO coefficient profiles of the 23 conventional and unconventional anthropometric indicators in pre-hypertension. A coefficient profile plot was produced against the log lambda sequence. In this study, anthropometric indicators were chosen according to the minimum criteria, where the optimal lambda resulted in 15 nonzero coefficients. (B) A 10-fold cross-validation was used in the LASSO regression for pre-hypertension. Binomial deviance was plotted versus log lambda. The dotted vertical lines were drawn at the optimal values using the minimum criteria (left dotted line) and the one standard error criteria (right dotted line). (C) LASSO coefficient profiles of the 24 conventional and unconventional anthropometric indicators in hypertension. A coefficient profile plot was produced against the log lambda sequence. In this study, anthropometric indicators were chosen according to the minimum criteria, where the optimal lambda resulted in 18 nonzero coefficients. (D) A 10-fold cross-validation was used in the LASSO regression for hypertension. Binomial deviance was plotted versus log lambda. The dotted vertical lines were drawn at the optimal values using the minimum criteria (left dotted line) and the one standard error criteria (right dotted line).

Association of Conventional and Unconventional Anthropometric Indicators with Pre-Hypertension and Hypertension

After excluded the variables with multi-collinearity, multivariate logistic regression analysis was used to analyze the association of conventional and unconventional anthropometric indicators with pre-hypertension and hypertension (Table 3). The final model adjusted for covariates such as demographic characteristics, health-related behaviors, family history, and blood biochemical indicators, and showed that weight (OR: 2.082, 95% CI: 1.280–3.411), BMI (OR: 4.047, 95% CI: 1.485–11.107), HtaW (OR: 0.998, 95% CI: 0.996–0.999), FMI (OR: 0.138, 95% CI: 0.034–0.556), FFMI (OR: 0.202, 95% CI: 0.061–0.661), AIP (OR: 0.331, 95% CI: 0.132–0.818), and TyG (OR: 2.467, 95% CI: 1.704–3.599) were significantly associated with pre-hypertension, while only FM (OR: 1.373, 95% CI: 1.004–1.895), AIP (OR: 0.115, 95% CI: 0.041–0.309), and TyG (OR: 5.450, 95% CI: 3.557–8.435) were significantly associated with hypertension.

Table 3 Multivariate Logistic Regression Analysis of Conventional and Unconventional Anthropometric Indicators with Pre-Hypertension and Hypertension

Discrimination Ability of Different Anthropometric Indicators by ROC Curves

Figure 3 showed the ROC curves for anthropometric indicators related to pre-hypertension and hypertension, and Table 4 demonstrated the area under curve (AUC), best threshold, sensitivity, and specificity of these anthropometric indicators. After adjusting for covariates, the predictive efficacy of the seven anthropometric indicators in pre-hypertension tended to be consistent, and the AUC of these anthropometric indicators were ranked from high to low as follows: Weight > BMI > FMI > FFMI > HtaW = TyG > AIP. Additionally, three anthropometric indicators had good predictive effects for hypertension, with their AUC ranked from high to low as follows: FM > TyG > AIP.

Table 4 AUCs and Best Thresholds for Anthropometric Indicators in Relation to Pre-Hypertension and Hypertension

Figure 3 ROC curves of anthropometric indicators for discriminating pre-hypertension (A) and hypertension (B). Notes: The ROC curves were adjusted for age, sex, current residence, educational level, annual household income per capita, occupation, medical insurance, household size, marital status, smoking, drinking, regular sleep, family history of hypertension, family history of cardiovascular diseases, FBG, TC, TG, HDL-C, LDL-C, UA (TG and HDL-C were not adjusted for AIP because they were included in the formula; TG and FBG were not adjusted for TyG because they were included in the formula).

Dose-Response Relationships of Anthropometric Indicators with Pre-Hypertension and Hypertension by RCS Models

According to Figure 4, RCS models suggested that weight, FMI, FFMI, AIP, and TyG had linear dose-response relationships with pre-hypertension risk (P for nonlinear > 0.05), while BMI and HtaW were nonlinearly associated with pre-hypertension risk (P for nonlinear < 0.05); FM, AIP and TyG had nonlinear dose-response relationships with hypertension risk (P for nonlinear < 0.05).

Figure 4 The restricted cubic splines of conventional and unconventional anthropometric indicators with the risk of pre-hypertension or hypertension. Notes: The restricted cubic spline analyses were adjusted for age, sex, current residence, educational level, annual household income per capita, occupation, medical insurance, household size, marital status, smoking, drinking, regular sleep, family history of hypertension, family history of cardiovascular diseases, FBG, TC, TG, HDL-C, LDL-C, UA (TG and HDL-C were not adjusted for AIP because they were included in the formula; TG and FBG were not adjusted for TyG because they were included in the formula). (A) dose-response relationship between weight and the risk of pre-hypertension; (B) dose-response relationship between BMI and the risk of pre-hypertension; (C) dose-response relationship between HtaW and the risk of pre-hypertension; (D) dose-response relationship between FMI and the risk of pre-hypertension; (E) dose-response relationship between FFMI and the risk of pre-hypertension; (F) dose-response relationship between AIP and the risk of pre-hypertension; (G) dose-response relationship between TyG and the risk of pre-hypertension; (H) dose-response relationship between FM and the risk of hypertension; (I) dose-response relationship between AIP and the risk of hypertension; (J) dose-response relationship between TyG and the risk of hypertension.

Subgroup Analyses and Sensitivity Analyses

Subgroup analyses revealed that among the conventional and unconventional anthropometric indicators of pre-hypertension, there were statistically significant interactions between age and weight, BMI, HtaW, FFMI, AIP; current residence and weight; educational level and weight, BMI, HtaW, FFMI, AIP, TyG; annual household income per capita, occupation, regular sleep and weight, BMI, HtaW, FFMI; medical insurance and weight, HtaW, FFMI; household size and weight, BMI, FFMI, AIP; marital status and weight, FFMI, AIP; smoking and HtaW, FMI, FFMI; drinking and BMI, FMI. Additionally, the interactions between age, occupation, family history of hypertension and AIP, TyG; sex, educational level, smoking and FM, AIP, TyG; medical insurance, marital status and AIP; household size and FM, AIP; drinking and FM were found to be statistically significant for hypertension (see Supplementary Figure 2). Furthermore, some sets of post hoc sensitivity analyses were performed to verify the robustness of the associations (see Supplementary Table 4). In most sensitivity analyses, the above associations between conventional and unconventional anthropometric indicators and pre-hypertension or hypertension remained unchanged.

Discussion

To the best of our knowledge, this is the first study to compare 24 indicators simultaneously in a large Chinese survey. We comprehensively assess the associations between a wide range of conventional and unconventional anthropometric indicators and pre-hypertension or hypertension in Chinese adults. In this study, we found that pre-hypertension was associated with weight, BMI, HtaW, FMI, FFMI, AIP, and TyG, with BMI showing the strongest association, followed by TyG. For hypertension, significant associations were found with FM, AIP, and TyG, with TyG exhibiting a notably stronger association compared to other anthropometric indicators. The ROC results demonstrated that conventional anthropometric indicators such as weight and BMI still had good predictive performance for pre-hypertension, while TyG and AIP, as unconventional anthropometric indicators, also have guiding significance to identify pre-hypertension. For hypertension, FM (a conventional anthropometric indicator), along with AIP and TyG (unconventional anthropometric indicators), showed excellent predictive values. The RCS models revealed that weight, FMI, FFMI, AIP, and TyG were linearly related to the risk of pre-hypertension, whereas BMI and HtaW were nonlinearly related to pre-hypertension risk; FM, AIP, and TyG were nonlinearly related to the risk of hypertension.

Weight, as one of the most fundamental conventional anthropometric indicators, serves as the most straightforward measure of obesity in individuals. Several cohort studies have demonstrated that weight change was associated with the incidence and mortality of metabolic diseases such as diabetes and hypertension,^39,40 as well as CVDs including coronary heart disease.⁴¹ In addition to being linked to the risk of disease in healthy populations, a prospective cohort study demonstrated that weight variability and weight change were both associated with higher risk of CVD mortality and all-cause mortality in individuals with hypertension.⁴² Another study also found that excessive weight can adversely affect kidney function through metabolic diseases.⁴³ This highlights the importance of daily weight management and monitoring individual weight changes, which are significant for both normotensive and hypertensive individuals. Although this study is a cross-sectional study, we still found that weight is an important predictor for pre-hypertension, suggesting that we should pay attention to our own weight and strive to maintain it within a normal range.

However, weight can only partially reflect overall obesity, and due to individual differences in height, using weight alone to predict hypertension risk is significantly biased. BMI and HtaW take into account the influence of height on weight, thereby partially addressing these shortcomings. Numerous cross-sectional studies have demonstrated that BMI was associated with pre-hypertension or hypertension.^44–46 A cohort study indicated that the risk of hypertension was not only related to the magnitude of BMI growth but also to the rate of growth, and that the risk of hypertension decreased significantly with the reduction in BMI.⁴⁷ Moreover, either a persistently high BMI or a rapid increase in BMI from childhood to adulthood may have adverse long-term effects on the development of hypertension and CVDs.⁴⁸ However, no studies have yet explored the relationship between HtaW and hypertension or pre-hypertension. Our study found that both BMI and HtaW have good predictive value for pre-hypertension, highlighting the importance of obesity indicators that consider height in the risk assessment of hypertension.

Although this study identified through ROC curve analysis that weight and BMI exhibit the highest predictive efficacy for pre-hypertension compared to other conventional anthropometric indicators, previous research indicated that BMI merely represented overall obesity without distinguishing between the proportions of muscle and fat in an individual’s weight, and this limitation gives rise to the obesity paradox, potentially introducing bias into the research outcomes.⁴⁹ FM refers to the total weight of all fat tissue within the body, typically measured using bioelectrical impedance analysis (BIA), dual-energy X-ray absorptiometry (DEXA), or skinfold thickness measurements. These methods provide a more accurate assessment of fat content, thereby evaluating individual obesity without the interference of muscle mass. FMI and FFMI, representing fat mass and fat-free mass adjusted for height, respectively, partially correct the bias caused by comparing FM across individuals of varying heights. Our study demonstrated that FMI and FFMI were predictive indicators for pre-hypertension risk, while FM was an indicator for hypertension risk, which aligns with the results of several other scholarly studies.^16,50,51 Besides, some researchers investigated the associations between FMI, FFMI and hypertension, but their findings have been inconsistent,^17,52,53 potentially due to differences in geographic regions, ethnicities, and study periods. However, our study does not support the predictive role of FMI and FFMI for hypertension risk. Future research should consider expanding the sample size to validate these findings in larger populations.

AIP is a biomarker composed of TG and HDL-C, which has been widely used in recent years to predict atherosclerosis and cardiovascular events, and has been demonstrated to have a higher predictive efficacy compared to individual lipid risk factors such as TG, TC, LDL-C, and HDL-C.³¹ Previous studies indicated that AIP was an important predictor of metabolic diseases such as hypertension, diabetes, and hyperuricemia,^54–56 and was associated with all-cause mortality and CVD-specific mortality in patients with hypertension.⁵⁷ The multivariate logistic regression results of this study showed that AIP was negatively correlated with the risk of pre-hypertension or hypertension, which is contrary to previous research.⁵⁸ However, the results of univariate logistic regression, RCS plots, and subgroup analyses indicated that AIP was positively correlated with the risk of pre-hypertension or hypertension. The possible reason is that despite controlling for multicollinearity among independent variables by screening variables through LASSO regression before conducting multivariate logistic regression, some multicollinearity or interaction among the selected variables still exists due to the intrinsic characteristics of anthropometric indicators, which led to a change in the association direction between AIP and pre-hypertension or hypertension. Further exploration is needed to investigate whether there are interactions between AIP and other anthropometric indicators. TyG is a recently popular “star” anthropometric indicator, combining TG and FBG, and has been identified as a reliable surrogate biomarker for insulin resistance.^34,59 Recent studies have also shown that it was closely related to the development and prognosis of CVDs.⁶⁰ Several scholars investigated the association between TyG and hypertension,^61,62 finding that TyG was an effective predictor of all-cause mortality in hypertensive patients.^63,64 However, no studies have yet focused on whether TyG is associated with pre-hypertension. Our study found that TyG was associated with both pre-hypertension and hypertension. Moreover, compared to AIP, TyG showed a stronger association and better predictive performance for pre-hypertension and hypertension, further confirming that TyG can serve as an effective predictor of the risk of pre-hypertension and hypertension.

Even if the diagnostic criteria for hypertension are well established, some scholars recently questioned the “one-size-fits-all” approach of only using blood pressure measurements to determine the presence of the condition. This is particularly significant for populous countries like China, where changes in the diagnostic threshold for hypertension could lead to substantial fluctuations in the number of diagnosed patients. Therefore, in the context of precision medicine, we should consider adopting more scientific and reasonable methods for individualized prediction of hypertension. Based on the above discussion, our study suggested that utilizing anthropometric indicators to predict the risk of pre-hypertension/hypertension could provide new insights for more accurate and scientific hypertension prediction in the future. We also attempt to propose a decision chart for individualized prediction of pre-hypertension/hypertension (Figure 5). In future primary care practice, screening strategies should be further refined, for example, by combining BMI with TyG testing to achieve more precise hypertension prediction.

Figure 5 Decision diagram for individualized prediction of pre-hypertension/hypertension.

This study innovatively explored the association of 24 conventional and unconventional anthropometric indicators with pre-hypertension or hypertension. By extensively adjusting for known confounding factors, we employed multiple methods to identify the optimal anthropometric indicators for predicting pre-hypertension or hypertension, which enriched the theoretical research on predictive factors for pre-hypertension and hypertension, providing scientific evidence and reasonable recommendations for reducing the incidence of hypertension. However, there are still some limitations in this study. Firstly, as a cross-sectional study, it cannot establish causal relationships. Secondly, despite our efforts to collect currently known anthropometric indicators through literature review, some new indicators might not have been measured. Thirdly, despite extensive covariate adjustment, lifestyle factors (eg, dietary intake, stress, salt consumption) were not captured and could influence blood pressure. Fourthly, due to the lack of hip circumference data, some related anthropometric indicators such as waist-to-hip ratio (WHR), abdominal volume index (AVI), and body adiposity index (BAI) could not be included in this study. We will further improve the survey design in future research. Lastly, it is undeniable that demographic characteristics may vary across different regions. Although our study subjects were selected through strict multi-stage stratified random sampling, it was limited to the population of Fujian Province. Therefore, caution should be exercised when generalizing the findings, and future studies should expand the sample size for further analysis.

In summary, the TyG index is emerging as a powerful marker for hypertension risk prediction and should be considered alongside conventional measures to strengthen primary care strategies.

Conclusions

In conclusion, this is one of the first comprehensive comparisons of 24 anthropometric indicators in a large Chinese population. We investigated the association of conventional and unconventional anthropometric indicators with the risk of pre-hypertension and hypertension in Chinese adults. Our results indicated that weight, BMI, HtaW, FMI, FFMI, AIP, and TyG were independently associated with pre-hypertension, among which the BMI and TyG had the strongest association with pre-hypertension, while hypertension was associated with FM, AIP, and TyG, with TyG showing a significantly stronger association with hypertension compared to other anthropometric indicators. Given the high prevalence of hypertension in China, simple conventional anthropometric measures still hold substantial potential for early population-level prevention. Emerging indicators such as TyG also deserve increased attention and could be integrated into existing clinical screening protocols to achieve more precise risk stratification, enabling clinicians to tailor lifestyle or therapeutic interventions accordingly. Future longitudinal studies are warranted to confirm causality and validate the predictive utility of novel indices such as TyG and AIP.

Abbreviations

CVD, cardiovascular disease; BMI, body mass index; WC, waist circumference; WHtR, waist-height ratio; BMR, basal metabolic rate; FM, fat mass; FFM, fat-free mass; FMI, fat mass index; FFMI, fat-free mass index; BFP, body fat percentage; VFI, visceral fat index; WCI, waist circumference index; WWI, weight-adjusted-waist index; BSA, body surface area; CI, conicity index; BRI, body roundness index; ABSI, a body shape index; AIP, atherogenic index of plasma; TyG, triglyceride-glucose index; TG, triglyceride; HDL-C, high-density lipoprotein cholesterol; FBG, fasting plasma glucose; LAP, lipid accumulation product; VAI, visceral adiposity index; CMI, cardiometabolic index; SBP, systolic blood pressure; DBP, diastolic blood pressure; TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; UA, uric acid; HtaW, height-adjusted weight; LASSO, least absolute shrinkage and selection operator; ROC, receiver-operating characteristic; RCS, restricted cubic spline; CNY, Chinese yuan; UEBMI, urban employees basic medical insurance; URRBMI, urban and rural residents basic medical insurance; AUC, area under curve; BIA, bioelectrical impedance analysis; DEXA, dual-energy x-ray absorptiometry; WHR, waist-to-hip ratio; AVI, abdominal volume index; BAI, body adiposity index.

Data Sharing Statement

The datasets used and/or analyzed during the current study are available from the corresponding author (Xian-E Peng) on reasonable request.

Ethics Approval and Consent to Participate

Ethics approval was obtained from the Ethics Committee of Fuwai Hospital (No. 2020-1360), and written informed consent was obtained from each participant. According to Item 2 of Article 32 of the Measures for Ethical Review of Life Science and Medical Research Involving Human Subjects dated February 18, 2023, our study is exempt from ethical review as it involves anonymized data (https://www.gov.cn/zhengce/zhengceku/2023-02/28/content_5743658.htm).

Acknowledgments

The authors would like to express their sincere gratitude to all the participants who wholeheartedly provided invaluable information and their collaboration in this research.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

This work was supported by the Project of the National Health Commission of the People’s Republic of China (grant numbers: NHC2020-609) and the Special Funding Project of Fujian Provincial Department of Finance (grant numbers: BPB-HY2021).

Disclosure

The authors report no conflicts of interest in this work.

References

1. Poulter NR, Prabhakaran D, Caulfield M. Hypertension. Lancet. 2015;386(9995):801–812. doi:10.1016/s0140-6736(14)61468-9

2. Han M, Li Q, Liu L, et al. Prehypertension and risk of cardiovascular diseases: a meta-analysis of 47 cohort studies. J Hypertens. 2019;37(12):2325–2332. doi:10.1097/hjh.0000000000002191

3. Zhou B, Perel P, Mensah GA, Ezzati M. Global epidemiology, health burden and effective interventions for elevated blood pressure and hypertension. Nat Rev Cardiol. 2021;18(11):785–802. doi:10.1038/s41569-021-00559-8

4. Mills KT, Stefanescu A, He J. The global epidemiology of hypertension. Nat Rev Nephrol. 2020;16(4):223–237. doi:10.1038/s41581-019-0244-2

5. NCD-RisC. Worldwide trends in blood pressure from 1975 to 2015: a pooled analysis of 1479 population-based measurement studies with 19·1 million participants. Lancet. 2017;389(10064):37–55. doi:10.1016/s0140-6736(16)31919-5

6. Clayton TL, Fitch A, Bays HE. Obesity and hypertension: obesity medicine association (OMA) clinical practice statement (CPS) 2023. Obes Pillars. 2023;8:100083. doi:10.1016/j.obpill.2023.100083

7. Dwivedi AK, Dubey P, Cistola DP, Reddy SY. Association between obesity and cardiovascular outcomes: updated evidence from meta-analysis studies. Curr Cardiol Rep. 2020;22(4):25. doi:10.1007/s11886-020-1273-y

8. Powell-Wiley TM, Poirier P, Burke LE, et al. Obesity and cardiovascular disease: a scientific statement from the American Heart Association. Circulation. 2021;143(21):e984–e1010. doi:10.1161/cir.0000000000000973

9. Kang NL. Association between obesity and blood pressure in common Korean people. Vasc Health Risk Manag. 2021;17:371–377. doi:10.2147/vhrm.S316108

10. Fan H, Guan T, Zhang X. Association of birthweight with overweight, obesity, and blood pressure among adolescents. Children. 2023;10(4):617. doi:10.3390/children10040617

11. Shojaei M, Jahromi AS, Karamatollah R. Association of obesity and pulse pressure with hypertension in an Iranian urban population. J Family Med Prim Care. 2020;9(9):4705–4711. doi:10.4103/jfmpc.jfmpc_723_20

12. Thapa J, Sundar Budhathoki S, Niraula SR, Pandey S, Thakur N, Pokharel PK. Prehypertension and its predictors among older adolescents: a cross-sectional study from eastern Nepal. PLOS Glob Public Health. 2022;2(9):e0001117. doi:10.1371/journal.pgph.0001117

13. Vo HK, Nguyen DV, Vu TT, Tran HB, Nguyen HTT. Prevalence and risk factors of prehypertension/hypertension among freshman students from the Vietnam National University: a cross-sectional study. BMC Public Health. 2023;23(1):1166. doi:10.1186/s12889-023-16118-4

14. Ononamadu CJ, Ezekwesili CN, Onyeukwu OF, Umeoguaju UF, Ezeigwe OC, Ihegboro GO. Comparative analysis of anthropometric indices of obesity as correlates and potential predictors of risk for hypertension and prehypertension in a population in Nigeria. Cardiovasc J Afr. 2017;28(2):92–99. doi:10.5830/cvja-2016-061

15. Ali N, Mahmood S, Manirujjaman M, et al. Hypertension prevalence and influence of basal metabolic rate on blood pressure among adult students in Bangladesh. BMC Public Health. 2017;18(1):58. doi:10.1186/s12889-017-4617-9

16. Ittermann T, Werner N, Lieb W, et al. Changes in fat mass and fat-free-mass are associated with incident hypertension in four population-based studies from Germany. Int J Cardiol. 2019;274:372–377. doi:10.1016/j.ijcard.2018.09.035

17. Takase M, Nakamura T, Tsuchiya N, et al. Association between the combined fat mass and fat-free mass index and hypertension: the Tohoku Medical Megabank Community-based Cohort Study. Clin Exp Hypertens. 2021;43(7):610–621. doi:10.1080/10641963.2021.1925681

18. Wang Z, Zeng X, Chen Z, et al. Association of visceral and total body fat with hypertension and prehypertension in a middle-aged Chinese population. J Hypertens. 2015;33(8):1555–1562. doi:10.1097/hjh.0000000000000602

19. Wang S, Peng R, Liang S, et al. Comparison of adiposity indices in relation to prehypertension by age and gender: a community-based survey in Henan, China. Clin Cardiol. 2018;41(12):1583–1592. doi:10.1002/clc.23086

20. Li Y, Gui J, Zhang X, et al. Predicting hypertension by obesity- and lipid-related indices in mid-aged and elderly Chinese: a nationwide cohort study from the China Health and Retirement Longitudinal Study. BMC Cardiovasc Disord. 2023;23(1):201. doi:10.1186/s12872-023-03232-9

21. Gao Q, Lin Y, Xu R, et al. Positive association of triglyceride-glucose index with new-onset hypertension among adults: a national cohort study in China. Cardiovasc Diabetol. 2023;22(1):58. doi:10.1186/s12933-023-01795-7

22. Xiao M, Chen C, Wang J, et al. Association of adiposity indices with prehypertension among Chinese adults: a cross-sectional study. J Clin Hypertens. 2023;25(5):470–479. doi:10.1111/jch.14622

23. James PA, Oparil S, Carter BL, et al. 2014 evidence-based guideline for the management of high blood pressure in adults: report from the panel members appointed to the Eighth Joint National Committee (JNC 8). JAMA. 2014;311(5):507–520. doi:10.1001/jama.2013.284427

24. WHO expert consultation. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet. 2004;363(9403):157–163. doi:10.1016/s0140-6736(03)15268-3

25. Gnatiuc L, Alegre-Díaz J, Halsey J, et al. Adiposity and blood pressure in 110 000 Mexican adults. Hypertension. 2017;69(4):608–614. doi:10.1161/hypertensionaha.116.08791

26. Ashwell M, Lejeune S, McPherson K. Ratio of waist circumference to height may be better indicator of need for weight management. BMJ. 1996;312(7027):377. doi:10.1136/bmj.312.7027.377

27. VanItallie TB, Yang MU, Heymsfield SB, Funk RC, Boileau RA. Height-normalized indices of the body’s fat-free mass and fat mass: potentially useful indicators of nutritional status. Am J Clin Nutr. 1990;52(6):953–959. doi:10.1093/ajcn/52.6.953

28. Hwaung P, Heo M, Kennedy S, et al. Optimum waist circumference-height indices for evaluating adult adiposity: an analytic review. Obes Rev. 2020;21(1):e12947. doi:10.1111/obr.12947

29. Park Y, Kim NH, Kwon TY, Kim SG. A novel adiposity index as an integrated predictor of cardiometabolic disease morbidity and mortality. Sci Rep. 2018;8(1):16753. doi:10.1038/s41598-018-35073-4

30. Mosteller RD. Simplified calculation of body-surface area. N Engl J Med. 1987;317(17):1098. doi:10.1056/nejm198710223171717

31. Fernández-Macías JC, Ochoa-Martínez AC, Varela-Silva JA, Pérez-Maldonado IN. Atherogenic index of plasma: novel predictive biomarker for cardiovascular illnesses. Arch Med Res. 2019;50(5):285–294. doi:10.1016/j.arcmed.2019.08.009

32. Kahn HS. The “lipid accumulation product” performs better than the body mass index for recognizing cardiovascular risk: a population-based comparison. BMC Cardiovasc Disord. 2005;5:26. doi:10.1186/1471-2261-5-26

33. Amato MC, Giordano C, Galia M, et al. Visceral Adiposity Index: a reliable indicator of visceral fat function associated with cardiometabolic risk. Diabetes Care. 2010;33(4):920–922. doi:10.2337/dc09-1825

34. Simental-Mendía LE, Rodríguez-Morán M, Guerrero-Romero F. The product of fasting glucose and triglycerides as surrogate for identifying insulin resistance in apparently healthy subjects. Metab Syndr Relat Disord. 2008;6(4):299–304. doi:10.1089/met.2008.0034

35. Valdez R. A simple model-based index of abdominal adiposity. J Clin Epidemiol. 1991;44(9):955–956. doi:10.1016/0895-4356(91)90059-i

36. Thomas DM, Bredlau C, Bosy-Westphal A, et al. Relationships between body roundness with body fat and visceral adipose tissue emerging from a new geometrical model. Obesity. 2013;21(11):2264–2271. doi:10.1002/oby.20408

37. Krakauer NY, Krakauer JC. A new body shape index predicts mortality hazard independently of body mass index. PLoS One. 2012;7(7):e39504. doi:10.1371/journal.pone.0039504

38. Wakabayashi I, Daimon T. The “cardiometabolic index” as a new marker determined by adiposity and blood lipids for discrimination of diabetes mellitus. Clin Chim Acta. 2015;438:274–278. doi:10.1016/j.cca.2014.08.042

39. Kim SH, Kwak JS, Kim SP, Choi SH, Yoon HJ. The association between diabetes and hypertension with the number and extent of weight cycles determined from 6 million participants. Sci Rep. 2022;12(1):5235. doi:10.1038/s41598-022-09221-w

40. Koebnick C, Sidell MA, Li X, Woolford SJ, Kuizon BD, Kunani P. Association of high normal body weight in youths with risk of hypertension. JAMA Network Open. 2023;6(3):e231987. doi:10.1001/jamanetworkopen.2023.1987

41. Lissner L, Odell PM, D’Agostino RB, et al. Variability of body weight and health outcomes in the Framingham population. N Engl J Med. 1991;324(26):1839–1844. doi:10.1056/nejm199106273242602

42. Kazibwe R, Singleton MJ, Ahmad MI, et al. Association between weight variability, weight change and clinical outcomes in hypertension. Am J Prev Cardiol. 2023;16:100610. doi:10.1016/j.ajpc.2023.100610

43. Fritz J, Brozek W, Concin H, et al. The association of excess body weight with risk of ESKD is mediated through insulin resistance, hypertension, and hyperuricemia. J Am Soc Nephrol. 2022;33(7):1377–1389. doi:10.1681/asn.2021091263

44. El-Agroudy AE, Arekat M, Jaradat A, et al. Pre-hypertension and hypertension among university students in Bahrain: a study of prevalence and associated risk factors. Cureus. 2024;16(3):e55989. doi:10.7759/cureus.55989

45. Hossain FB, Adhikary G, Chowdhury AB, Shawon MSR. Association between body mass index (BMI) and hypertension in south Asian population: evidence from nationally-representative surveys. Clin Hypertens. 2019;25:28. doi:10.1186/s40885-019-0134-8

46. Wang Y, Min C, Song X, et al. The dose-response relationship between BMI and hypertension based on restricted cubic spline functions in children and adolescents: a cross-sectional study. Front Public Health. 2022;10:870568. doi:10.3389/fpubh.2022.870568

47. Guo B, Shi Z, Zhang W, et al. Trajectories of body mass index (BMI) and hypertension risk among middle-aged and elderly Chinese people. J Hum Hypertens. 2021;35(6):537–545. doi:10.1038/s41371-020-0368-7

48. Twig G, Yaniv G, Levine H, et al. Body-mass index in 2.3 million adolescents and cardiovascular death in adulthood. N Engl J Med. 2016;374(25):2430–2440. doi:10.1056/NEJMoa1503840

49. Zhou Q, Liu X, Zhao Y, et al. BMI and risk of all-cause mortality in normotensive and hypertensive adults: the rural Chinese cohort study. Public Health Nutr. 2021;24(17):5805–5814. doi:10.1017/s1368980021001592

50. Khaleghi MM, Jamshidi A, Afrashteh S, et al. The association of body composition and fat distribution with hypertension in community-dwelling older adults: the Bushehr Elderly Health (BEH) program. BMC Public Health. 2023;23(1):2001. doi:10.1186/s12889-023-16950-8

51. Li M, Lin J, Liang S, Huang S, Wen Z, Mo Z. Predicted fat mass, lean body mass, and risk of hypertension: results from a Chinese male cohort study. Obes Facts. 2022;15(5):638–647. doi:10.1159/000524653

52. Abolhasani M, Maghbouli N, Karbalai Saleh S, et al. Which anthropometric and metabolic index is superior in hypertension prediction among overweight/obese adults? Integr Blood Press Control. 2021;14:153–161. doi:10.2147/ibpc.S340664

53. Rao KM, Arlappa N, Radhika MS, Balakrishna N, Laxmaiah A, Brahmam GN. Correlation of fat mass index and fat-free mass index with percentage body fat and their association with hypertension among urban South Indian adult men and women. Ann Hum Biol. 2012;39(1):54–58. doi:10.3109/03014460.2011.637513

54. Yuan Y, Shi J, Sun W, Kong X. The positive association between the atherogenic index of plasma and the risk of new-onset hypertension: a nationwide cohort study in China. Clin Exp Hypertens. 2024;46(1):2303999. doi:10.1080/10641963.2024.2303999

55. Li Y, Zeng L. Comparison of seven anthropometric indexes to predict hypertension plus hyperuricemia among U.S. adults. Front Endocrinol. 2024;15:1301543. doi:10.3389/fendo.2024.1301543

56. Li YW, Kao TW, Chang PK, Chen WL, Wu LW. Atherogenic index of plasma as predictors for metabolic syndrome, hypertension and diabetes mellitus in Taiwan citizens: a 9-year longitudinal study. Sci Rep. 2021;11(1):9900. doi:10.1038/s41598-021-89307-z

57. Duiyimuhan G, Maimaiti N. The association between atherogenic index of plasma and all-cause mortality and cardiovascular disease-specific mortality in hypertension patients: a retrospective cohort study of NHANES. BMC Cardiovasc Disord. 2023;23(1):452. doi:10.1186/s12872-023-03451-0

58. Tan M, Zhang Y, Jin L, et al. Association between atherogenic index of plasma and prehypertension or hypertension among normoglycemia subjects in a Japan population: a cross-sectional study. Lipids Health Dis. 2023;22(1):87. doi:10.1186/s12944-023-01853-9

59. Ramdas Nayak VK, Satheesh P, Shenoy MT, Kalra S. Triglyceride Glucose (TyG) Index: a surrogate biomarker of insulin resistance. J Pak Med Assoc. 2022;72(5):986–988. doi:10.47391/jpma.22-63

60. Tao LC, Xu JN, Wang TT, Hua F, Li JJ. Triglyceride-glucose index as a marker in cardiovascular diseases: landscape and limitations. Cardiovasc Diabetol. 2022;21(1):68. doi:10.1186/s12933-022-01511-x

61. Zheng R, Mao Y. Triglyceride and glucose (TyG) index as a predictor of incident hypertension: a 9-year longitudinal population-based study. Lipids Health Dis. 2017;16(1):175. doi:10.1186/s12944-017-0562-y

62. Xin F, He S, Zhou Y, Jia X, Zhao Y, Zhao H. The triglyceride glucose index trajectory is associated with hypertension: a retrospective longitudinal cohort study. Cardiovasc Diabetol. 2023;22(1):347. doi:10.1186/s12933-023-02087-w

63. Pang J, Qian L, Che X, Lv P, Xu Q. TyG index is a predictor of all-cause mortality during the long-term follow-up in middle-aged and elderly with hypertension. Clin Exp Hypertens. 2023;45(1):2272581. doi:10.1080/10641963.2023.2272581

64. Zhou D, Liu XC, Kenneth L, Huang YQ, Feng YQ. a non-linear association of triglyceride glycemic index with cardiovascular and all-cause mortality among patients with hypertension. Front Cardiovasc Med. 2021;8:778038. doi:10.3389/fcvm.2021.778038

LONDON — LONDON (AP) — Amazon says a massive outage of its cloud computing service has been resolved as of Monday evening, after a problem disrupted internet use around the world, taking down a broad range of online services, including social media, gaming, food delivery, streaming and financial platforms.

The all-day disruption and the ensuing exasperation it caused served as the latest reminder that 21st century society is increasingly dependent on just a handful of companies for much of its internet technology, which seems to work reliably until it suddenly breaks down.

About three hours after the outage began early Monday morning, Amazon Web Services said it was starting to recover, but it wasn’t until 6 p.m. Eastern that “services returned to normal operations,” Amazon said on its AWS health website, where it tracks outages.

AWS provides behind-the-scenes cloud computing infrastructure to some of the world’s biggest organizations. Its customers include government departments, universities and businesses, including The Associated Press.

Cybersecurity expert Mike Chapple said “a slow and bumpy recovery process” is “entirely normal.”

As engineers roll out fixes across the cloud computing infrastructure, the process could trigger smaller disruptions, he said.

“It’s similar to what happens after a large-scale power outage: While a city’s power is coming back online, neighborhoods may see intermittent glitches as crews finish the repairs,” said Chapple, an information technology professor at the University of Notre Dame’s Mendoza College of Business.

Amazon pinned the outage on issues related to its domain name system that converts web addresses into IP addresses, which are numeric designations that identify locations on the internet. Those addresses allow websites and apps to load on internet-connected devices.

DownDetector, a website that tracks online outages, said in a Facebook post that it received over 11 million user reports of problems at more than 2,500 companies. Users reported trouble with the social media site Snapchat, the Roblox and Fortnite video games, the online broker Robinhood and the McDonald’s app, as well as Netflix, Disney+ and many other services.

The cryptocurrency exchange Coinbase and the Signal chat app both said on X that they were experiencing trouble related to the outage.

Amazon’s own services were also affected. Users of the company’s Ring doorbell cameras and Alexa-powered smart speakers reported that they were not working, while others said they were unable to access the Amazon website or download books to their Kindle.

Many college and K-12 students were unable to submit or access their homework or course materials Monday because the AWS outage knocked out Canvas, a widely used educational platform.

“I currently can’t grade any online assignments, and my students can’t access their online materials” because of the outage’s effect on learning-management systems, said Damien P. Williams, a professor of philosophy and data science at the University of North Carolina at Charlotte.

The exact number of schools impacted was not immediately known, but Canvas says on its website it is used by 50% of college and university students in North America, including all Ivy League schools in the U.S.

At the University of California, Riverside, students couldn’t submit assignments, take quizzes or access course materials, and online instruction was limited, the campus said.

Ohio State University informed its 70,000 students at all six campuses by email Monday morning that online course materials might be inaccessible due to the outage and that “students should connect with their instructors for any alternative plans.” As of 7:10 p.m. Eastern, access was restored, the university told students.

This is not the first time issues with Amazon cloud services have caused widespread disruptions.

Many popular internet services were affected by a brief outage in 2023. AWS’s longest outage in recent history occurred in late 2021, when a wide range of companies — from airlines and auto dealerships to payment apps and video streaming services — were affected for more than five hours. Outages also happened in 2020 and 2017.

The first signs of trouble emerged at around 3:11 a.m. Eastern time, when AWS reported on its “health dashboard” that it was “investigating increased error rates and latencies for multiple AWS services in the US-EAST-1 Region.” Later, the company reported that there were “significant error rates” and that engineers were “actively working” on the problem.

Around 6 a.m. Eastern time, the company reported seeing recovery across most of the affected services and said it was seeking a “full resolution.” As of midday, AWS was still working to resolve the trouble.

Sixty-four internal AWS services were affected, the company said.

Because much of the world now relies on three or four companies to provide the underlying infrastructure of the internet, “when there’s an issue like this, it can be really impactful” across many online services, said Patrick Burgess, a cybersecurity expert at U.K.-based BCS, The Chartered Institute for IT.

“The world now runs on the cloud,” Burgess said.

And because so much of the online world’s plumbing is underpinned by so few companies, when something goes wrong, “it’s very difficult for users to pinpoint what is happening because we don’t see Amazon, we just see Snapchat or Roblox,” Burgess said.

“The good news is that this kind of issue is usually relatively fast” to resolve, and there’s no indication that it was caused by a cyberattack, Burgess said.

“This looks like a good old-fashioned technology issue. Something’s gone wrong, and it will be fixed by Amazon,” he said.

There are “well-established processes” to deal with outages at AWS, as well as rivals Google and Microsoft, Burgess said, adding that such outages are usually over in “hours rather than days.”

___

Ortutay reported from San Francisco. Associated Press videojournalist Mustakim Hasnath in London and Jocelyn Gecker in San Francisco contributed to this report.

Introduction

Pancreatic cancer is one of the most aggressive malignancies, with only 20% of patients eligible for surgical resection at the time of diagnosis.^1–3 These patients often face prolonged hospitalization and significant postoperative challenges, among which pain control remains a major clinical concern. Poorly managed postoperative pain can stimulate catecholamine release, which may suppress natural killer cell activity—a component of innate immunity—and potentially influence anti-tumor responses.⁴ Additionally, it is associated with increased psychological distress and reduced quality of life. Despite its clinical significance, current pain management strategies after pancreatic surgery are often suboptimal, underscoring the need for more effective analgesic approaches and further investigation into their impact on postoperative outcomes.

Patient-controlled intravenous analgesia (PCIA) with opioids is widely used for postoperative pain control.^5–8 Sufentanil, a selective potent μ-receptor agonist, is widely used for its efficacy in postoperative pain management.⁹ Given the moderate to severe pain typically associated with pancreatic surgery, a potent analgesic strategy is essential. However, increasing the dosage of a single analgesic agent to achieve adequate pain relief may also elevate the risk of adverse effects, including respiratory depression, nausea, and vomiting. Dezocine, a partial μ-receptor agonist and κ-receptor antagonist, has emerged as a promising adjunct due to its analgesic and sedative effects, as well as its favorable safety profile compared to pure μ-receptor agonists.^10–12 By acting on κ-receptors in the spinal cord and brain, dezocine provides analgesic and sedative effects without the typical µ-receptor dependence, potentially reducing adverse reactions such as smooth muscle relaxation.¹⁰ Previous studies have demonstrated that dezocine offers significant postoperative antihyperalgesic and analgesic effects, with benefits lasting up to 48 hours in patients undergoing open gastrectomy.¹³

Several studies have demonstrated that dezocine, when combined with morphine, enhances postoperative analgesia and reduces opioid-related side effects, such as nausea and pruritus, making it a valuable option in anesthesia practice.^14–16 At our institution, the combination of sufentanil and dezocine has been used in PCIA for pancreatic cancer patients for several years. However, the efficacy and safety of this combination have not been thoroughly investigated. To address this gap, we conducted a propensity score-matched (PSM) study at a high-volume pancreatic center to evaluate the role of dezocine as an adjunct to sufentanil in PCIA for postoperative pain management following pancreatic surgery, which, to our knowledge, is the first study to investigate the analgesic effects of this combination in PCIA for pancreatic surgery patients.

Materials and Methods

Ethics Approval

This retrospective study was approved by the Ethics Committee of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine (Ethics Approval Number: (2023) No. 48), with a waiver of patient written informed consent due to the use of de-identified, archival medical records (no active patient intervention). All patient identifiers were removed, and data were stored securely on encrypted servers accessible only to the research team, adhering to the Declaration of Helsinki (as revised in 2013).

Patients

A total of 1485 patients who underwent elective open or minimally invasive pancreatic tumor surgery and received patient-controlled intravenous analgesia (PCIA) for postoperative pain management at the Pancreas Center of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, between January 2022 and January 2023 were retrospectively enrolled. The center is one of the largest pancreatic surgery centers in Asia. Among them, 794 were male and 691 were female, with an age range of 18 to 85 years (mean age: 60.55 ± 12.55 years) and American Society of Anesthesiologists (ASA) physical status classification ranging from I to IV. Based on the PCIA regimen, patients were allocated into two groups: the sufentanil group (n = 251) and the sufentanil-dezocine combination group (n = 1234). Surgical approach (Laparotomy/Laparoscopic/Robotic) was documented based on the description of the surgical procedure in the operative notes. All operative notes were reviewed and signed off by the attending surgeon or a senior resident physician to ensure consistency in classification. To minimize confounding and selection bias, PSM was performed using a logistic regression model based on age, sex, BMI, surgical approach (laparotomy, laparoscopic, robotic), surgery type (pancreatoduodenectomy, total pancreatectomy, middle-preserving pancreatectomy, distal pancreatectomy, as different techniques may affect pain severity due to varying tissue trauma), and dexmedetomidine dose. A caliper of 0.02 and nearest-neighbor matching were applied in a 1:3 ratio using R software (v.4.3.1, The R Foundation for Statistical Computing, Vienna, Austria. http://www.r-project.org). Exclusion criteria included: (1) known allergies to study drugs; (2) inability to use patient-controlled intravenous analgesia (PCIA); (3) history of chronic pain or long-term use of analgesic medications; (4) requirement for reoperation due to postoperative bleeding or severe abdominal infection; (5) severe cardiopulmonary or hepatorenal insufficiency and (6) cognitive dysfunction.

Anesthesia Procedure

All patients fasted for 8 hours (solids) and 6 hours (clear liquids) preoperatively and were transferred to the operating room without premedication. Standard monitoring included electrocardiography (ECG), non-invasive blood pressure (BP), respiratory rate (RR), oxygen saturation (SpO₂), end-tidal carbon dioxide pressure (PetCO₂), and bispectral index (BIS). A uniform anesthetic regimen was administered to all patients, with surgeries performed by the same surgical team.

General anesthesia was induced with propofol (2–2.5mg/kg), sufentanil (0.3–0.5 µg/kg), rocuronium (0.6–0.8mg/kg) or cisatracurium (0.2–0.3mg/kg), dexamethasone 5 mg, and dexmedetomidine 0.6 µg/kg. Preoxygenation with 100% oxygen was administered for at least 3 minutes via a face mask. Anesthesia was maintained with sevoflurane (3vol%, 0.8–1.3MAC), remifentanil (0.2–0.4µg/kg/min), supplemental rocuronium (1/3–1/5 of the induction dose), and intermittent sufentanil (0.4 µg/kg). Ventilation was set at a tidal volume of 8 mL/kg, with respiratory frequency adjusted to maintain PetCO₂ at 35–45 mmHg. Anesthesia depth was titrated to maintain a BIS between 40 and 60, ensuring mean arterial pressure (MAP) and heart rate (HR) remained within 20% of baseline values. Patient temperature was maintained above 36°C using infusion heaters and warming blankets. A sufentanil loading dose (0.1 µg/kg) was administered 30 minutes before the end of surgery. Intraoperative fluid balance was defined as the net change in a patient’s total body fluid volume during surgery, calculated as the difference between the total intraoperative fluids inputs and outputs. Postoperatively, patients were transferred to the post-anesthesia care unit (PACU), where residual neuromuscular blockade was reversed with neostigmine (40 µg/kg) and atropine (20 µg/kg).

Postoperative PCIA Regimen

After meeting extubation criteria, patients were extubated and connected to an Artificial Intelligence Patient-Controlled Analgesia (AI-PCA) system (Model ZZB-IB, Nantong AIPU Medical Inc., China). Patients were divided into two groups based on the PCIA solution: the sufentanil group received sufentanil (1.0 µg/mL), and the combination group received sufentanil (1.0 µg/mL) plus dezocine (2.5 mg/mL). Group allocation was guided by clinical judgment of the anesthesiologist considering factors reflected in our dataset such as patient demographics, surgical complexity, intraoperative management details.

The Acute Pain Service team prepared the PCIA solution in 100 mL normal saline bags, containing either sufentanil alone or the combination and monitored patients. If the Numerical Rating Scale (NRS) at rest was ≥4, a 2 mL bolus of PCIA solution was administered at 15-minute intervals until NRS <4. Patients were then encouraged to self-administer PCIA as needed.

The PCIA pump was set to a background infusion rate of 2 mL/h, with a 2 mL bolus dose and a 15-minute lockout interval. PCIA was maintained for 48 hours postoperatively, during which vital signs including respiratory rate, oxygen saturation, and sedation scores were closely monitored.

Outcome Measures

Demographic and intraoperative data, including surgery type, site, anesthetic drug dosages, blood loss, transfusion, and fluid balance, were recorded. Postoperative data included PCIA pump usage duration, total input, cumulative and effective press counts, rescue analgesia, and adverse events (eg, vomiting, pruritus, respiratory depression, hypotension, dizziness, delirium). We assessed Functional Activity Score (FAS) and Level of Sedation (LOS) at 1 and 2 days post-surgery. FAS (1–3 grades) quantifies pain impact on daily functions: 1=no limitation (normal coughing/limb movement despite pain); 2=mild limitation (slight difficulty/slower actions), and 3=severe limitation (struggles with basic activities). LOS (0–3 grades) evaluates consciousness via responsiveness: 0=alert (follows instructions), 1=somnolent (wakes to calls but drifts off), 2=stuporous (brief pain wakefulness), 3=comatose (no response to calls or pain). Both were graded during routine checks to guide pain management and monitor recovery. Pain intensity was evaluated using the NRS at rest (NRSR) and during coughing (NRSC) at 24, 48, and 72 hours post-surgery. The NRS ranges from 0 (no pain) to 10 (worst imaginable pain). Moderate-to-severe pain was defined as NRS ≥4. Mild pain (NRS 1–3) was also recorded in postoperative data. Adverse events were recorded based on routine clinical documentation in the hospital’s electronic medical records (EMR) and nursing care logs.

Primary endpoints were the incidence of moderate-to-severe pain at rest and during coughing within 48 hours post-surgery. Secondary endpoints included the incidence of moderate-severe pain at rest and during coughing at 24 hours and 72 hours post-surgery, LOS, FAS, and adverse events.

Statistical Analysis

Continuous variables were first assessed for normality, those with normal distribution were expressed as mean ± standard deviation (SD) and compared using independent t-tests. Skewed distributed continuous variables were presented as median (Q1, Q3) and analyzed with the Mann–Whitney U-test. Categorical variables were expressed as frequencies and percentage, and compared using Pearson’s chi-square or Fisher’s exact test. Missing data for demographic characteristics, intraoperative and postoperative data were imputed using the expectation-maximization algorithm. Univariate and multivariate logistic regression models, alongside with post-PSM analysis and inverse probability weighting (IPW) analysis were conducted to calculate odds ratios (OR) and 95% confidence intervals (CI). Analyses were performed using SAS (v.9.2, SAS Institute Inc., USA). All tests were two-sided, and statistical significance was set at the 5% level. No adjustments have been made for multiple testing.

Results

Patient Characteristics

Before PSM, the sufentanil group comprised 251 patients, while the combination group included 1234 patients. The sufentanil group was older (mean age 63.73 ± 13.69 years vs 59.90 ± 13.44 years, P< 0.05), had a higher proportion of pancreatoduodenectomy (PD) procedures (55.8% vs 44.8%, P< 0.05), and a greater rate of laparotomy (80.5% vs 73.9%, P < 0.05). Additionally, the sufentanil group had a lower BMI (22.25 ± 3.30 vs 22.73 ± 3.31, P < 0.05) and received a lower dexmedetomidine dosage (16.85 ± 15.75 µg vs 22.50 ± 16.32 µg, P < 0.05) compared to the combination group. No significant difference was observed in sex distribution. After PSM, the study included 247 patients in the sufentanil group and 704 in the combination group, with all baseline variables balanced between the two groups (Table 1).

Table 1 Demographic Characteristics and Perioperative Outcomes of Patients Between the Sufentanial Group and the Combination Group

Perioperative Outcomes

After PSM, no significant differences were observed in blood loss, blood transfusion volume, or total PCIA input between the two groups, despite differences before matching. The dosages of sufentanil and rocuronium bromide, as well as effective and cumulative PCIA press counts, showed no significant differences before or after PSM. However, the sufentanil group exhibited greater fluid balance difference and longer pump usage duration, which were statistically significant both before and after PSM (Table 1).

Primary Endpoint

The incidence of moderate-to-severe pain at rest and during coughing within 48 hours post-surgery is summarized in Table 2. After PSM, 19 patients (7.7%) in the sufentanil group experienced moderate-to-severe pain at rest, compared to 20 patients (2.8%) in the combination group (P < 0.05). Similarly, the incidence of pain during coughing was significantly higher in the sufentanil group (74 patients, 30.0%) than in the combination group (166 patients, 23.6%) during the same period (P < 0.05). These differences were also observed before PSM.

Table 2 Moderate-Severe Pain at Rest and During Coughing After Surgery Between the Sufentanial Group and the Combination Group

At 48 hours post-surgery, NRSR was significantly higher in the sufentanil group (1.97 ± 1.26) compared to the combination group (1.77 ± 0.91) (P= 0.018). Similarly, NRSC at 48 hours was higher in the sufentanil group (3.13 ± 1.57) than in the combination group (2.89 ± 1.17) (P= 0.022). All four analytical approaches including univariate and multivariate logistic regression analyses, post- PSM analysis and IPW analysis consistently identified sufentanil monotherapy as an independent predictor of moderate-to-severe pain, with odds ratios (ORs) and 95% confidence intervals (CIs) presented in Table 3.

Table 3 Logistic Regression Results for Moderate-Severe Pain at Rest and During Coughing After Surgery Between the Sufentanial Group and the Combination Group

Secondary Endpoints

Significant differences in the incidence of pain at rest and during coughing were observed at 24 and 72 hours post-surgery before PSM (P < 0.05). After PSM, these differences remained significant, except for pain during coughing at 72 hours (Table 2). No significant inter-group differences were noted in vomiting, hypotension, dizziness, delirium, or rescue analgesia on the first and second postoperative days, either before or after PSM. However, the functional activity scale (FAS) scores on the first and second postoperative days revealed significant differences between the two groups. Additionally, the proportion of fully alert patients on the second postoperative day was significantly higher in the combination group compared to the sufentanil group, both before and after PSM (Table 4).

Table 4 Adverse Events Between the Sufentanial Group and the Combination Group

Discussion

Pancreatic surgery is a critical intervention for pancreatic cancer, yet patients often experience prolonged postoperative pain, which can hinder physical and mental recovery. Effective pain management is therefore essential for improving patient outcomes and has garnered significant clinical attention. Opioid-based analgesia, particularly sufentanil, is widely used in patient-controlled intravenous analgesia (PCIA). However, the adverse effects of opioids, such as addiction, respiratory depression, pruritus, and sedation, have driven the search for alternative strategies to reduce opioid dosages and minimize side effects.¹⁷ Multimodal analgesia has emerged as a promising approach.¹⁸

In this propensity score-matched study, we evaluated the efficacy of combining sufentanil with dezocine in PCIA for postoperative pain management in patients undergoing pancreatic surgery. After matching, baseline characteristics and perioperative outcomes were comparable between the groups. Our findings demonstrated that the sufentanil-dezocine combination significantly reduced the incidence of moderate-to-severe pain at rest and during coughing within the first 48 hours postoperatively, without increasing the risk of clinically relevant side effects such as vomiting, hypotension, dizziness, delirium, or the need for rescue analgesia. Patients in the combination group exhibited significantly lower NRSR and NRSC at 48 hours post-surgery compared to the sufentanil group. Multivariate logistic regression analysis identified sufentanil monotherapy as an independent predictor of postoperative pain, suggesting that the addition of dezocine enhances analgesic efficacy, consistent with previous findings.¹⁶ These findings align with dezocine’s proposed mechanism: by targeting κ-receptors (which modulate pain perception) and partially activating μ-receptors (avoiding overstimulation), the combination may enhance analgesia while mitigating pure μ-agonist-related side effects. Notably, the reduction in pain during coughing–a high-pain activity critical for pancreatic surgery recovery–suggests the combination may be particularly beneficial for patients requiring early mobilization.

A primary concern with combining dezocine and sufentanil in PCIA is the potential for excessive sedation. However, our study found no evidence of increased sedation in the combination group during the 48-hour postoperative period. While sedation levels on the first postoperative day did not differ significantly, the proportion of fully alert patients was significantly higher in the combination group on the second postoperative day. This finding suggests that dezocine may enhance patient alertness while maintaining effective analgesia–Its ability to improve alertness and reduce sedation-related complications supports its value as a “balanced” adjunct in postoperative pain management^19–21 likely due to κ-receptor activation inducing lighter sedation compared to μ-agonists.

Postoperative adverse events, such as vomiting, hypotension, and dizziness, can negatively impact patient satisfaction and prolong hospital stays.¹⁰ Our study found that the addition of dezocine to sufentanil did not exacerbate these side effects. Notably, the combination group had a significantly lower incidence of respiratory depression compared to the sufentanil group, with no significant differences in vomiting, hypotension, dizziness, or delirium. These results align with previous research^22–25 and further support the safety profile of the sufentanil-dezocine combination.

Despite these promising findings, several limitations should be acknowledged. First, the retrospective design of the study introduces potential for selection bias, although this was mitigated by propensity score matching and the uniformity of our surgical team, Sensitivity analyses using alternative matching strategies (eg inverse probability weighting, multivariate logistic regression) yielded consistent results, suggesting no major residual confounding affected our conclusions. Second, generalizability of our findings may be limited. Due to our single-center design, even though our cohort meets high-volume criteria. As emphasized in a recent review on gastric cancer surgery outcomes, institutional factors can create variability in textbook outcomes (TOs) even among high-volume centers, highlighting the need for cross-institutional validation.²⁶ Future multi-center collaborations will compare textbook outcomes across 10+ high-volume centers using a pragmatic, standardized protocol to address this gap. Third, retrospective data precluded optimization of sufentanil/dezocine dosing. Prospective dose-response studies are needed to refine postoperative pain management in high-risk surgical populations.

In conclusion, our study demonstrates that the sufentanil-dezocine combination in PCIA significantly reduces moderate-to-severe pain at rest and during coughing within the first 48 hours after pancreatic surgery, without increasing the incidence of clinically relevant adverse effects, which has not been previously reported in the context of pancreatic surgery, suggesting it as a promising and safe approach for postoperative pain management in pancreatic cancer patients. Future research should focus on optimizing dosing strategies and confirming these results in prospective, multicenter trials.

Data Sharing Statement

The original contributions presented in the study are included in the article; further inquiries can be directed to the corresponding authors.

Funding

This work was supported by the National Natural Science Foundation of China (No: T2293734).

Disclosure

The authors declare that they have no conflicts of interest in this work.

References

1. Zhang Q, Zeng L, Chen Y, et al. Pancreatic cancer epidemiology, detection, and management. Gastroenterol Res Pract. 2016;2016:10. doi:10.1155/2016/8962321

2. van Dijk SM, Heerkens HD, Tseng DSJ, et al. Systematic review on the impact of pancreatoduodenectomy on quality of life in patients with pancreatic cancer. HPB. 2018;20(3):204–215. doi:10.1016/j.hpb.2017.11.002

3. Karamarković AR, Juloski JT. Current surgical concepts and future perspectives in the treatment of borderline resectable and potentially resectable locally advanced pancreatic cancer. Chirurgia. 2022;117(4):385–398. doi:10.21614/chirurgia.2770

4. Min EK, Chong JU, Hwang HK. Negative oncologic impact of poor postoperative pain control in left-sided pancreatic cancer. World J Gastroenterol. 2017;23(4):676–686. doi:10.3748/wjg.v23.i4.676

5. Song JW, Shim JK, Song Y, et al. Effect of ketamine as an adjunct to intravenous patient-controlled analgesia, in patients at high risk of postoperative nausea and vomiting undergoing lumbar spinal surgery. Br J Anaesth. 2013;111:630–635. doi:10.1093/bja/aet192

6. Klotz R, Larmann J, Klose C, et al. Gastrointestinal complications after pancreatoduodenectomy with epidural vs patient-controlled intravenous analgesia: a randomized clinical trial. JAMA Surg. 2020;155(7):e200794. doi:10.1001/jamasurg.2020.0794

7. Gostian M, Loeser J, Bentley T, et al. Analgesia after tonsillectomy with controlled intravenous morphine – overdue or exaggerated? Braz J Otorhinolaryngol. 2023;89(1):48–53. doi:10.1016/j.bjorl.2021.08.002

8. Liu F, Li TT, Yin L, et al. Analgesic effects of sufentanil in combination with flurbiprofen axetil and dexmedetomidine after open gastrointestinal tumor surgery: a retrospective study. BMC Anesthesiol. 2022;22(1):130. doi:10.1186/s12871-022-01670-0

9. Lindemann C, Strube P, Fisahn C, et al. Patient-controlled sublingual sufentanil tablet system versus intravenous opioid analgesia for postoperative pain management after lumbar spinal fusion surgery. Eur Spine J. 2023;32(1):321–328. doi:10.1007/s00586-022-07462-x

10. Zhu H, Chen YB, Huang SQ, et al. Interaction of analgesic effects of dezocine and sufentanil for relief of postoperative pain: a pilot study. Drug Des Devel Ther. 2020;14:4717–4724. doi:10.2147/DDDT.S270478

11. Xia Y, Sun Y, Liu J. Effects of dezocine on PAED scale and Ramsay sedation scores in patients undergoing NUSS procedure. Am J Transl Res. 2021;13(5):5468–5475.

12. Ye RR, Jiang S, Xu X, et al. Dezocine as a potent analgesic: overview of its pharmacological characterization. Acta Pharmacol Sin. 2022;43(7):1646–1657. doi:10.1038/s41401-021-00790-6

13. Yu F, Zhou J, Xia S, et al. Dezocine prevents postoperative hyperalgesia in patients undergoing open abdominal surgery. Evid Based Complement Alternat Med. 2015;2015:946194. doi:10.1155/2015/946194

14. Sun ZT, Yang CY, Cui Z, et al. Effect of intravenous dezocine on fentanyl-induced cough during general anesthesia induction: a double-blinded, prospective, randomized, controlled trial. J Anesth. 2011;25:860–863. doi:10.1007/s00540-011-1237-x

15. Zhu Y, Jing G, Yuan W. Preoperative administration of intramuscular dezocine reduces postoperative pain for laparoscopic cholecystectomy. J Biomed Res. 2011;25:356–361. doi:10.1016/S1674-8301(11)60047-X

16. Wu L, Dong YP, Sun L, Sun L. Low concentration of dezocine in combination with morphine enhance the postoperative analgesia for thoracotomy. J Cardiothorac Vasc Anesth. 2015;29(4):950–954. doi:10.1053/j.jvca.2014.08.012

17. Li QZ, Yao HX, Xu MY, et al. Dedetomidine combined with sufentanil and dezocine-based patient controlled intravenous analgesia increases female patients’ global satisfaction degree after thoracoscopic surgery. J Cardiothorac Surg. 2021;16(1):102. doi:10.1186/s13019-021-01472-4

18. Gritsenko K, Khelemsky Y, Kaye AD, et al. Multimodal therapy in perioperative analgesia. Best Pract Res Clin Anaesthesiol. 2014;28(1):59–79. doi:10.1016/j.bpa.2014.03.001

19. Barr GA, Schmidt HD, Thakrar AP, Kranzler HR, Liu R. Revisiting dezocine for opioid use disorder: a narrative review of its potential abuse liability. CNS Neurosci Ther. 2024;30(9):e70034. doi:10.1111/cns.70034

20. Schmidt HD, Zhang Y, Xi J, et al. A new formulation of dezocine, Cycdezocine, reduces oxycodone self-administration in female and male rats. Neurosci Lett. 2023;815:137479. doi:10.1016/j.neulet.2023.137479

21. Grothusen J, Lin W, Xi J, et al. Dezocine is a biased ligand without significant beta-arrestin activation of the mu opioid receptor. Transl Perioper Pain Med. 2022;9(1):424–429.

22. Wang CY, Li LZ, Shen BX, et al. A multicenter randomized double-blind prospective study of the postoperative patient controlled intravenous analgesia effects of dezocine in elderly patients. Int J Clin Exp Med. 2014;7(3):530–539.

23. He LX, Yao YT, Shao K, et al. Efficacy of dezocine on preventing opioid-induced cough during general anaesthesia induction: a PRISMA-compliant systematic review and meta-analysis. BMJ Open. 2022;12(4):e052142. doi:10.1136/bmjopen-2021-052142

24. Zhang L, Li C, Zhao C, et al. Analgesic comparison of dezocine plus propofol versus fentanyl plus propofol for gastrointestinal endoscopy: a meta-analysis. Medicine. 2021;100(15):e25531. doi:10.1097/MD.0000000000025531

25. Gui YK, Zeng XH, Xiao R, et al. The Effect of dezocine on the median effective dose of sufentanil-induced respiratory depression in patients undergoing spinal anesthesia combined with low-dose dexmedetomidine. Drug Des Devel Ther. 2023;17:3687–3696. doi:10.2147/DDDT.S429752

26. Marano L, Verre L, Carbone L, et al. Current trends in volume and surgical outcomes in gastric cancer. J Clin Med. 2023;12(7):2708. doi:10.3390/jcm12072708

Introduction

Hepatocellular carcinoma (HCC) is one of the most prevalent and lethal malignancies worldwide, with hepatectomy remaining a primary curative treatment option for early-stage disease.^1,2 However, postoperative recurrence poses a significant challenge, with 5-year recurrence rate as high as 60%.³ Recurrence can be categorized as early recurrence (<2 years), caused by micro metastases after resection, or late recurrence (>2 years), caused by new tumors arising in a microenvironment prone to carcinogenesis.⁴ Research has shown that HCC patients who experience early recurrence post-resection have a worse prognosis compared to those with late recurrence.⁵ Therefore, accurate prediction of early recurrence is critical for personalized postoperative surveillance and adjuvant therapy strategies.

Certain clinicopathological features, including the Barcelona Clinic Liver Cancer (BCLC) staging system, alpha-fetoprotein (AFP), microvascular invasion (MVI), and satellite nodules (SNs), have been linked to postoperative prognosis.^1,6 However, conventional predictive assessments based on these biomarkers lack sufficient predictive accuracy, and information on MVI and SNs is not available preoperatively. Several recent studies have used artificial intelligence models based on CT/MRI data to effectively predict the risk of postsurgical HCC recurrence risk.^7–10 Compared with CT and MRI, ultrasound possesses the advantages of real-time, radiation-free, and cost-effective, and can also be used for intraoperative evaluation and guide the operation pathway. Additionally, contrast-enhanced ultrasound (CEUS) has emerged as a valuable imaging modality for HCC, offering real-time dynamic assessment of tumor vascularity and perfusion characteristics.¹¹

Previous studies have reported that certain CEUS characteristics such as rapid wash-in, early wash-out phase, rapid portal phase signal regression, and high LI-RADS grading are risk factors associated with early recurrence after radical HCC resection.^12,13 However, a critical gap remains: the interpretation of these characteristics is highly subjective and difficult to quantify, which limits their consistent application in prognostication. To address this gap in objectivity, deep learning (DL) emerges as a powerful tool. DL algorithms can automatically extract massive, quantitative features from medical images to model disease diagnosis and prognosis.¹⁴ The application of DL to liver CEUS is already established for tasks such as focal liver lesion classification and predicting HCC biological behavior,^15,16 demonstrating its potential in this domain. However, these existing applications primarily focus on classification limited to single-phase analysis or static images, potentially overlooking the critical time-dependent kinetic information. The specific knowledge gap our study aims to fill is the use of DL to predict long-term clinical outcomes directly from pre-operative dynamic CEUS cines. Therefore, we hypothesize that a DL-based framework, by leveraging the rich temporal data in multiphase CEUS, can overcome the limitations of subjective interpretation. This study aims to develop and validate such a framework, integrating CEUS data with clinicopathological variables to provide a noninvasive and efficient preoperative tool for predicting early recurrence in patients with early-stage HCC after surgical resection.

Methods

Study Population and Design

This study was performed in accordance with the principles of the Declaration of Helsinki. This retrospective study was approved by the Ethics Committee of the Second Affiliated Hospital of Nanchang University (Approval No: IIT-O-2025-251). The requirement for informed consent was waived by the Ethics Committee due to the retrospective and minimal-risk nature of the study. All patient data were kept strictly confidential and de-identified prior to analysis. From January 2010 to January 2023, 155 patients with HCC who underwent CEUS within two weeks before hepatectomy were enrolled. The diagnosis of HCC was based on the diagnostic criteria established by the European Association for the Study of the Liver.² Inclusion criteria were: BCLC stage 0 or A; liver function with Child-Pugh class A or B; performance status Eastern Cooperative Oncology Group score 0 or 1; R0 resection;¹⁷ regular follow-up within two years after hepatectomy. The exclusion criteria were poor CEUS imaging quality, recurrent HCC or primary HCC combined with other primary tumors, and a history of antitumor treatment before surgery, such as radiofrequency ablation, microwave ablation, transcatheter arterial chemoembolization, or chemotherapy. Pretreatment baseline clinical characteristics, including demographic data, laboratory test results, and clinical diagnoses, were collected from the Institutional Picture Archiving and Communication System (PACS^®; Carestream Health, Toronto, Canada). Finally, 115 patients’ preoperative CEUS cines and clinical variables were retrospectively analyzed. We developed four CEUS-based AI models to preoperatively predict early recurrence: CEUS-AP, CEUS-PP, CEUS-LP, and CEUS-MP. We further integrated clinical variables into the CEUS AI model to construct an individualized nomogram for preoperative prediction of early recurrence. The patient enrollment workflow and study design were shown in Figure 1.

Figure 1 Study workflow flowchart. Abbreviations: CEUS, contrast-enhanced ultrasound; HCC, hepatocellular carcinoma; ROI annotation illustration; AP, arterial phase; PP, portal venous phase; LP, late phase; MP, multiple phases.

CEUS Data Acquisition

CEUS was performed by one of two radiologists (ZH and YC) with over 10 years of experience in liver CEUS. Two types of ultrasound instruments were used in this study, Toshiba Aplio (n = 98) and SuperSonic Imagine (n = 17). The system parameters, including gain, dynamic range, mechanical index, output power, and focal zone, were fine-tuned to ensure effective tissue signal reduction without compromising penetration capability. A volume of 2.4 mL of the second-generation contrast agent (SonoVue^®; Bracco Imaging, Milan, Italy) was injected within one second via the elbow, followed by a 5-mL saline flush. The transducer was maintained in a stable position to continuously observe perfusion of the lesion. Three-phase contrast enhancement was dynamically monitored, consisting of arterial phase (AP, 0–30 s), portal venous phase (PP, 31–120 s), and late phase (LP, 121–360 s). The CEUS digital video recordings were stored as DICOM format.

Hepatectomy Procedure and Follow-Up Protocol

Anatomical partial hepatectomy was performed with a resection margin of ≥ 10 mm. All enrolled tumors were safely and completely removed, with no residual tumors following resection (R0 classification). The patients underwent regular follow-up assessments at 1, 3, 6, 9, and 12 months postoperatively, followed by evaluations every 3–6 months thereafter. During these visits, contrast-enhanced imaging (CEUS, CECT, or CEMRI) and serum biomarker testing (AFP and liver function) were performed. In this study, early recurrence was defined as a new lesion with typical imaging features of HCC that appeared in imaging examinations within 2 years after hepatectomy or as confirmed by pathological results from percutaneous liver biopsy or a second surgery.

Developing Prediction Model Based on CEUS

This study employed CEUS cines from three phases (arterial phase, portal venous phase, and late phase) to construct the corresponding single-phase models (CEUS-AP, CEUS-PP, and CEUS-LP). We then developed a comprehensive CEUS-MP (multi-phase) model that integrated all three phases. In each patient, three separate regions of interest (ROIs) were outlined by a radiologist with 5-year experience in CEUS to enable quantitative assessment of the CEUS videos across the arterial, portal venous, and late phases. The annotation tool was ITK-SNAP, which is an open-source software.¹⁸ Specifically, the doctor first annotated the contour of the tumor in the CEUS frame when the tumor displayed clearly. A rectangle to cover the tumor contour and an approximately 0.5 cm wide liver parenchyma surrounding the tumor were created (Figure 2). All frames with the same ROI location determined in the annotated frame were cropped and manually adjusted if necessary. To assess inter-observer variability in ROI delineation, two doctors (one with more than 10 years of experience in CEUS and the other with more than 5 years of experience) independently contoured the ROIs for a randomly selected cohort of 30 patients, blinded to patient outcomes and clinical characteristics. Radiomic features extracted from both sets of ROIs were compared using the intraclass correlation coefficient (ICC) to determine feature reproducibility.

Figure 2 ROI annotation illustration. (a–c) An example of the red/green/blue rectangle ROI segmented in one AP/PP/LP CEUS frame when the tumor displayed clearly. Abbreviations: ROI, region of interest; AP, arterial phase; PP, portal venous phase; LP, late phase; CEUS, contrast-enhanced ultrasound.

Enrolled patients were randomly divided into training (n = 75) and validation (n = 40) cohorts. To minimize model-induced bias, convolutional neural networks (CNN) were employed for the four models based on CEUS cines in different phases consisting of a two-dimensional (2D) convolution layer, 2D max-pooling layer, ReLU activation function, vector of locally aggregated descriptors (VLAD), and a fully connecting layer (FCL) (Figure 3). The inputs of the CEUS-AP, CEUS-PP, CEUS-LP, and CEUS-MP models were different CEUS cines in the arterial, portal venous, late, and the above-mentioned three phases, respectively. In the four CEUS AI models, convolution layers automatically learn the imaging features from each frame of CEUS, and VLAD is responsible for the quantitative analysis of enhancement patterns over time for the CEUS cines. The FCL guide model predicted the probability of early recurrence for each patient. A ten-fold validation method was adopted in the training phase to optimize the four CEUS AI models. To improve the prediction performance and reduce the risk of overfitting, CEUS cines were split over time into multiple independent samples at two frames per second; in this way, the training population was augmented for better learning. Data augmentation techniques, including random rotations, flips, and intensity variations, were employed to augment the diversity of the training dataset. Moreover, depth-wise convolution was selected to reduce the number of parameters in the training phase. Both L1 and L2 regularization were employed to combat overfitting. L1 regularization promoted sparsity and can perform feature selection by driving less important weights to zero, while L2 regularization penalized large weights to ensure the model was less sensitive to small fluctuations in the input data. This combined approach was chosen to create a more robust and generalizable model given the high-dimensional feature space relative to our sample size. We implemented early stopping based on the validation loss to halt training once performance on the validation set ceased to improve. In this study, the convolutional layers employed a depthwise separable convolution structure to replace the conventional 3D convolution operations, separating the processes of feature convolution filtering and the integration of cross-channel features. A global pooling layer was incorporated before the fully connected layers, aggregating features across temporal and spatial dimensions to reduce the number of neural nodes that the fully connected layers need to process. Through a series of specialized network design strategies, the parameter count of R-DLCEUS was reduced to approximately 20,000. During the training phase, the model used cross-entropy as the objective function and optimized the parameters using the stochastic gradient descent method. Each batch consisted of 16 samples, and the model was trained for 70 epochs. The initial learning rate was set to 0.001 and was reduced by a factor of 0.5 at the 15th, 30th, and 55th epochs. A momentum coefficient of 0.9 was applied during training. For the CEUS-MP model, the single-channel cine loops from the AP, PP, and VP were concatenated along the channel dimension to form a multi-channel input, which was then processed by the subsequent 2D convolutional network. The area under the receiver operating characteristic curve (AUC) was adopted to quantitatively measure the prediction performance of the CEUS AI models.

Figure 3 The overall network structure of CEUS AI models. CNN were employed for the four models based on CEUS cines in different phases. Input step: The CEUS cines of AP/ PP/LP/MP were inputted into the CNN model, respectively. Feature extraction step: The features extracted from three phases were then integrated into one feature collection which demonstrated the characteristics of the entire CEUS cine. 2D Conv x included 2D convolution layer, 2D max-pooling layer, and ReLU activation function. Output step: probability was calculated to estimate risk of early recurrence after hepatectomy. Abbreviations: CEUS, contrast-enhanced ultrasound; AI, artificial intelligence; CNN, convolutional neural network; AP, arterial phase; PP, portal phase; LP, late phase; MP, multiple phases; 2D Conv, two-dimensional convolution; VLAD, vector of locally aggregated descriptors; FCL, full connected layer.

Building Combined Individualized Nomogram

To incorporate information from clinical variables with CEUS cines, an individualized model was developed to preoperatively predict early recurrence after hepatectomy. Specifically, four models developed from CEUS video data (CEUS-AP, CEUS-PP, CEUS-LP, and CEUS-MP) were evaluated to identify the optimal predictive model. Subsequently, the predictive probability of the selected model was used as a Radiomics signature and integrated with clinical variables to build a combined nomogram prediction model based on the multi-variable logistic regression analysis. In order to demonstrate the incremental value of DL-CEUS-based nomogram, univariate and multivariate logistic regression analyses were performed to build the Clinical-only model. The analyzed clinical variables included 16 basic clinical characteristics (Table 1). Variable selection followed a two-stage process. First, all parameters showing marginal significance (P < 0.10) in univariate analyses were included; subsequently, these were incorporated into a stepwise multivariable regression model. Only variables demonstrating statistical significance (P < 0.05) in the final multivariable analysis were retained as independent predictors and were used for combined nomogram development. The prediction performance of the nomogram was assessed using AUC, calibration curves, and decision curves.

Table 1 Baseline Characteristics of Patients

Visualization of the CEUS-MP Model

To enhance the interpretability of the DL model’s predictions regarding early recurrence, we employed Selvaraju R.’s method to transform the DL feature maps into pseudo-colored visualization maps.¹⁹ In these maps, warm-colored (red) pixels represent regions with stronger predictive relevance, indicating high-weight features that significantly contribute to the model’s output. Conversely, the cool-colored (blue) pixels denote areas of weaker correlation corresponding to the low-weight factors in the prediction. This visualization approach effectively highlights the critical image regions that influence the model’s decision-making process.

Statistical Analysis

The statistical software and packages used were Python (version 3.11), PyTorch (version 2.0.1), R (version 3.4.4), and computeC. The chi-square test was used to compare categorical variables. Student’s t-test or the Mann–Whitney test, as appropriate, was used to compare continuous variables. All statistical tests were two-sided, and differences were considered significant at P < 0.05.

Results

Clinical Characteristics

Among 115 patients, there were 93 males (93/115, 80.9%) and 22 females (22/115, 19.1%). The mean tumor size was 3.1 ± 0.9 cm (range, 1.0–5.0 cm). Postoperative follow-up results revealed that among the115 patients, 30 (30/115, 26.1%) experienced early recurrence, whereas 85 (85/115, 73.9%) showed no evidence of early recurrence. There were no significant differences in the baseline characteristics between the training (n = 75) and validation (n = 40) cohorts. Detailed baseline characteristics of the two cohorts are shown in Table 1.

Predictive Performance of CEUS AI Models

The features demonstrated high reproducibility, with 91% achieving excellent ICC values, indicating strong inter-observer agreement in ROI segmentation. In the training cohort, AUCs of CEUS-MP, CEUS-AP, CEUS-PP, and CEUS-LP reached 0.922 (95% CI: 0.816–0.971), 0.829 (95% CI: 0.712–0.954), 0.816 (95% CI: 0.697–0.941) and 0.808 (95% CI: 0.691–0.938), respectively. The DL-based AI model using multiple CEUS cines (CEUS-MP) achieved the best prediction performance compared with the other three models based on single-phase CEUS cine (CEUS-AP, CEUS-PP, and CEUS-LP) (Figure 4a). Similar results were observed in the validation cohort. In the validation cohort, the AUCs of CEUS-MP, CEUS-AP, CEUS-PP, and CEUS-LP were 0.840 (95% CI: 0.701–0.989), 0.741 (95% CI: 0.670–0.921), 0.719 (95% CI: 0.637–0.909), and 0.703 (95% CI: 0.661–0.889), respectively (Figure 4b). Table 2 provides an overview of the quantitative assessment of the four models. CEUS-MP offered significantly higher AUCs than CEUS-AP, CEUS-PP, and CEUS-LP in both training (P = 0.026, P = 0.014, P = 0.001) and validation (P = 0.029, P = 0.016, P = 0.001) cohorts. However, no significant differences in AUCs were observed between CEUS-AP, CEUS-PP, and CEUS-LP in either the training or validation cohorts (all P > 0.05).

Table 2 Predictive Performance of CEUS-MP, CEUS-AP, CEUS-PP, and CEUS-LP in Training and Validation Cohorts

Figure 4 ROC curves of four CEUS AI models. (a) ROC curves of CEUS AI models in training cohort. (b) ROC curves of CEUS AI models in validation cohort. CEUS-MP achieved the best prediction performance compared with CEUS-AP, CEUS-PP, and CEUS-LP in both cohorts.

The Individualized Predictive Value of Combined Nomogram

Multivariable regression analysis revealed that serum albumin (< 35 g/L) (HR = 1.146, P = 0.034), AFP (> 1000 ng/mL) (HR = 1.908, P = 0.028), and the CEUS-MP signature (HR = 5.141, P < 0.001) were independent preoperative predictors of early recurrence (Table 3). An individualized nomogram was built based on these three significant values (Figure 5a). In the training and validation cohorts, the AUCs of the nomogram for early recurrence was 0.945 (95% CI: 0.861–0.965) and 0.871 (95% CI: 0.751–0.970) for early recurrence, respectively. (Figure 5b). The nomogram achieved a sensitivity of 88.9% (95% CI: 75.9–96.1%) and specificity of 98.1% (95% CI: 81.6–99.1%) in the training cohort. In the validation cohort, the sensitivity and specificity of the nomogram were 83.0% (95% CI: 70.4–90.8%) and 82.5% (95% CI: 70.8–91.5%), respectively. The nomogram showed good calibration for early recurrence prediction (Figure 5c). The Hosmer-Lemeshow test showed that there were no significant differences in early recurrence prediction between the nomogram and the ideal reference curve (P = 0.415). The decision curve analysis of the nomogram is presented in Figure 5d. AFP and serum albumin were selected to construct a Clinical-only model. The Clinical-only model had an AUC of 0.573 (95% CI: 0.501–0.675) in the training cohort with a 0.554 (95% CI: 0.512–0.685) accuracy and an AUC of 0.528 (95% CI: 0.500–0.618) in the validation cohort with a 0.545 (95% CI: 0.509–0.678) accuracy. The nomogram achieved a significantly higher predictive performance compared to the Clinical-only model in both training and validation cohorts (Both P < 0.001). The decision curves showed that if the threshold probability of a patient was > 30%, integrating CEUS-MP signatures and clinical variables (AFP and serum albumin) to predict the patient’s early recurrence added more net benefit than using only clinical variables. In addition, the net benefit of the CEUS-MP signature was similar to that of the nomograms.

Table 3 Prognostic Factors to Predict Early Recurrence of HCC Patients After Hepatectomy

Figure 5 Prediction performance of individualized nomogram, ROC curves, calibration curves and Decision curve analysis. (a) Nomogram for predicting the probability of a patient occur early recurrence after hepatectomy. A patient who obtained high total scores of the three factors tends to have high probability of early recurrence after hepatectomy. ALB, 0 means < 35 g/L, 1 means ≥ 35 g/L; AFP, 0 means < 200ng/ mL, 1 means ≥ 200ng/mL. (b) ROC curves of nomogram in training and validation cohort. (c) Calibration curves of the nomogram for predicting early recurrence in training and validation cohorts. (d) Decision curve analysis for nomogram.

Understanding the R-DLCEUS Model

From the visualization maps of the CEUS-LP model, it can be observed that the model focuses more on the early period AP, suggesting its potential ability to capture the “fast-in” feature, which may be closely associated with the early recurrence of HCC. Additionally, we observed that for some early cases, the attention of the model exhibited a “patchy” pattern (Figure 6). We hypothesize that these heterogeneous attention regions may correspond to areas of necrotic or fibrotic tissue, features often associated with aggressive tumor biology. However, this requires further pathological correlation.

Figure 6 Visualization of the CEUS-LP model. (a–c) Representative time-point images (9s, 12s, 15s; early period AP) from dynamic CEUS in a post-hepatectomy patient with early recurrence. Each set of images displayed, from left to right, the original monochrome ultrasound image, the pseudo-color heatmap, and the heatmap overlaid on the original ultrasound image.

Discussion

This study developed a DL-based model using CEUS to preoperatively predict early recurrence in patients with early-stage HCC patients following hepatectomy. Our results demonstrated that the multiphase CEUS model (CEUS-MP) achieved superior predictive performance compared with single-phase models, with AUC of 0.922 and 0.840 in the training and validation cohorts, respectively. Furthermore, the integration of CEUS-MP signatures with clinical variables (serum albumin and AFP) into a nomogram further enhanced predictive accuracy, yielding AUCs of 0.945 and 0.871 in the training and validation cohorts, respectively. These findings highlighted the potential of CEUS combined with DL as an auxiliary method for preoperative risk stratification in patients with HCC.

While conventional ultrasound remains the first-line imaging tool for liver screening, its inability to assess the tumor vasculature limits its diagnostic and prognostic utility. Since CEUS cines overcome these limitations by providing dynamic, contrast-enhanced visualization of tumor microvasculature, leading to more accurate HCC diagnosis, recurrence prediction, and treatment monitoring.²⁰ Given these advantages, we utilized CEUS cines, rather than conventional ultrasound images, to develop our prediction models in this study, with the help of advanced quantification techniques for more precise analysis. The superior performance of the CEUS-MP model highlighted the importance of leveraging multiphase CEUS data, which capture dynamic tumor perfusion characteristics across the arterial, portal venous, and late phases. This aligns with previous studies emphasizing the prognostic value of CEUS features, such as rapid wash-in and early wash-out, in HCC recurrence prediction.¹² However, traditional CEUS interpretation is limited by subjectivity and inter-observer variability. Our DL approach addressed this limitation by automatically extracting quantitative features from CEUS cines, thereby standardizing the analysis and improving the reproducibility. Huang et al conducted a quantitative analysis of CEUS images of patients with hepatocellular carcinoma (HCC) following radical resection to predict early recurrence.²¹ However, their model achieved an AUC of only 0.57 in the testing cohort, which was significantly lower than that of our proposed model. A potential limitation of their approach was the reliance on a single-frame analysis, the peak contrast intensity of the lesion on CEUS, which may have decreased predictive accuracy. By contrast, our study leveraged continuous multiphase CEUS cine imaging, enabling a more comprehensive assessment and significantly improving the precision of early recurrence prediction.

The integration of CEUS-MP signatures with clinical variables into a nomogram represented a significant advancement in personalized HCC management. Multivariate analysis identified serum albumin and AFP as the independent clinical prognostic predictors. Notably, low ALB levels served as a reliable indicator of advanced hepatic dysfunction, reflecting both the progression of liver carcinogenesis and deteriorating hepatic synthetic capacity.²² Concurrently, elevated AFP served as a biological marker indicating high invasiveness in HCC.²³ In such patients, some microlesions may escape detection by imaging examinations during the initial HCC diagnosis, contributing to early recurrence after surgical resection. This study demonstrated that higher AFP levels were correlated with an increased risk of early tumor recurrence. Therefore, the AFP level held a significant value in predicting the prognosis of therapy. The high sensitivity and specificity of the nomogram in both cohorts suggested its clinical utility in identifying high-risk patients who may benefit from other treatment selections or intensified surveillance. Notably, the decision curve analysis indicated that the nomogram provided a greater net benefit than the clinical variables alone when the threshold probability exceeded 30%, supporting its practical application in clinical decision-making. The visualization maps further enhanced interpretability, revealing that the model focused on regions with heterogeneous enhancement patterns, which may correlate with aggressive tumor biology. The heterogeneous or peripheral enhancement patterns highlighted by the model were often associated with active tumor regions, while the lack of enhancement in certain areas may correlate with central necrosis or fibrosis commonly found in larger lesions.

Limitations

Our study has several limitations. First, it is a retrospective study conducted at a single center, which introduces potential selection bias. Our patient cohort may not fully represent the more heterogeneous patient populations seen across different healthcare institutions. Second, the uniform high-standard scanning protocols and specialized operators may have inflated the performance estimates of our model. Moreover, the relatively small cohort, particularly the limited number of early recurrence events, is a recognized limitation that warrants caution in interpreting the findings and underscores the need for future validation in larger populations. Furthermore, the use of two different ultrasound systems for CEUS acquisition, despite our efforts at intensity normalization, represents a potential technical confounder. Therefore, the generalizability of our findings needs to be further validated in real-world, multi-center settings, and should also explore incorporating additional imaging modalities (eg, MRI or CT) and molecular biomarkers to refine predictive accuracy.

Conclusion

In conclusion, our study demonstrates that a DL-based CEUS framework, particularly when combined with clinical variables, shows strong potential for the preoperative prediction of early recurrence in patients with HCC undergoing hepatectomy. While these initial results are promising, further validation in large-scale, multi-center, prospective cohorts is essential to confirm its generalizability and establish clinical utility. If validated, this approach could facilitate personalized postoperative management and potentially improve long-term patient outcomes. Future work should also focus on integrating this model with other imaging modalities, such as CT or MRI, to further enhance predictive accuracy.

Ethics Approval and Consent to Participate

This study was approved by the Institutional Review Board of The Second Affiliated Hospital of Nanchang University.

Acknowledgments

The authors thank the data collectors for their efforts and interest in participating in data collection. We would like to thank the patients who willingly provided all the necessary information without any reservation.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

This study was supported by grants from the National Natural Science Foundation of China (Grant No. 82360348), the Jiangxi Provincial Natural Science Foundation (Grant No. 20232BAB216096), and the Jiangxi Ganpo Outstanding Talent Support Program–Academic and Technical Discipline Leader Development Project (Grant No. 20243BCE51175).

Disclosure

All authors have no conflicts of interest to declare in this work.

References

1. Forner A, Reig M, Bruix J. Hepatocellular carcinoma. Lancet. 2018;391(10127):1301–1314. doi:10.1016/S0140-6736(18)30010-2

2. European Association for the Study of the Liver. EASL clinical practice guidelines: management of hepatocellular carcinoma. J Hepatol. 2018;69(1):182–236. doi:10.1016/j.jhep.2018.03.019

3. Lim KC, Chow PK, Allen JC, et al. Systematic review of outcomes of liver resection for early hepatocellular carcinoma within the Milan criteria. Br J Surg. 2012;99:1622–1629. doi:10.1002/bjs.8915

4. Llovet JM, Kelley RK, Villanueva A, et al. Hepatocellular carcinoma. Nat Rev Dis Primers. 2021;7(1):6. doi:10.1038/s41572-020-00240-3

5. Poon RT, Fan ST, Ng IO, et al. Different risk factors and prognosis for early and late intrahepatic recurrence after resection of hepatocellular carcinoma. Cancer. 2000;89(3):500–507.

6. Villanueva A. Hepatocellular carcinoma. N Engl J Med. 2019;380(15):1450–1462. doi:10.1056/NEJMra1713263

7. Lee IC, Huang JY, Chen TC, et al. Evolutionary learning-derived clinical-radiomic models for predicting early recurrence of hepatocellular carcinoma after resection. Liver Cancer. 2021;10(6):572–582. doi:10.1159/000518728

8. Kucukkaya AS, Zeevi T, Chai NX, et al. Predicting tumor recurrence on baseline MR imaging in patients with early-stage hepatocellular carcinoma using deep machine learning. Sci Rep. 2023;13(1):7579. doi:10.1038/s41598-023-34439-7

9. Yuan C, Wang Z, Gu D, et al. Prediction early recurrence of hepatocellular carcinoma eligible for curative ablation using a Radiomics nomogram. Cancer Imaging. 2019;19(1):21. doi:10.1186/s40644-019-0207-7

10. Hui TCH, Chuah TK, Low HM, et al. Predicting early recurrence of hepatocellular carcinoma with texture analysis of preoperative MRI: a radiomics study. Clin Radiol. 2018;73(12):1056.e11–1056.e16. doi:10.1016/j.crad.2018.07.109

11. Alzaraa A, Gravante G, Chung WY, et al. Contrast-enhanced ultrasound in the preoperative, intraoperative and postoperative assessment of liver lesions. Hepatol Res. 2013;43(8):809–819. doi:10.1111/hepr.12044

12. Cao K, Wu L, Wang X, et al. Risk factors for early recurrence after radical resection of hepatocellular carcinoma based on preoperative contrast-enhanced ultrasound and clinical features. Technol Cancer Res Treat. 2024;23:15330338241281327. doi:10.1177/15330338241281327

13. Terzi E, Iavarone M, Pompili M, et al. Contrast ultrasound LI-RADS LR-5 identifies hepatocellular carcinoma in cirrhosis in a multicenter restropective study of 1,006 nodules. J Hepatol. 2018;68(3):485–492. doi:10.1016/j.jhep.2017.11.007

14. Wang K, Lu X, Zhou H, et al. Deep learning radiomics of shear wave elastography significantly improved diagnostic performance for assessing liver fibrosis in chronic hepatitis B: a prospective multicenter study. Gut. 2019;68(4):729–741. doi:10.1136/gutjnl-2018-316204

15. Ding W, Meng Y, Ma J, et al. Contrast-enhanced ultrasound-based AI model for multi-classification of focal liver lesions. J Hepatol. 2025;83(2):426–439. doi:10.1016/j.jhep.2025.01.011

16. Xu W, Zhang H, Zhang R, et al. Deep learning model based on contrast-enhanced ultrasound for predicting vessels encapsulating tumor clusters in hepatocellular carcinoma. Eur Radiol. 2025;35(2):989–1000. doi:10.1007/s00330-024-10985-0

17. Hermanek P, Wittekind C. Residual tumor (R) classification and prognosis. Semin Surg Oncol. 1994;10(1):12–20. doi:10.1002/ssu.2980100105

18. Yushkevich PA, Piven J, Hazlett HC, et al. User-guided 3D active contour segmentation of anatomical structures: significantly improved efficiency and reliability. Neuroimage. 2006;31(3):1116–1128. doi:10.1016/j.neuroimage.2006.01.015

19. Selvaraju RR, Cogswell M, Das A. Grad-CAM: visual explanations from deep networks via gradient-based localization. In: Proceedings of the IEEE International Conference on Computer Vision. 2017:618–626.

20. Kudo M, Ueshima K, Osaki Y, et al. B-Mode ultrasonography versus contrast-enhanced ultrasonography for surveillance of hepatocellular carcinoma: a prospective multicenter randomized controlled trial. Liver Cancer. 2019;8(4):271–280. doi:10.1159/000501082

21. Huang Z, Shu Z, Zhu RH, et al. Deep learning-based radiomics based on contrast-enhanced ultrasound predicts early recurrence and survival outcome in hepatocellular carcinoma. World J Gastrointest Oncol. 2022;14(12):2380–2392. doi:10.4251/wjgo.v14.i12.2380

22. Huang YH, Wu JC, Chen CH, et al. Comparison of recurrence after hepatic resection in patients with hepatitis B vs. hepatitis C-related small hepatocellular carcinoma in hepatitis B virus endemic area. Liver Int. 2005;25(2):236–241. doi:10.1111/j.1478-3231.2005.01081.x

23. Yang L, Gu D, Wei J, et al. A radiomics nomogram for preoperative prediction of microvascular invasion in hepatocellular carcinoma. Liver Cancer. 2019;8(5):373–386. doi:10.1159/000494099

Introduction

Globally, hepatocellular carcinoma (HCC) remains a leading cause of cancer-related mortality, with increasing incidence projected over the next two decades. In China, it is the second most lethal cancer, and the long-term prognosis following curative surgery remains poor due to high recurrence rates, with 5-year recurrence rates reported at approximately 50%–70%.^1–6 Recurrence is most frequent within the first postoperative year and often appears as distant intrahepatic or extrahepatic metastases, thought to originate from undetected micrometastases present at surgery. Tumor-related features such as size, number, differentiation, microvascular invasion (MVI), and elevated alpha-fetoprotein (AFP) are major determinants of early recurrence. Conversely, factors including patient age, sex, underlying liver disease etiology, and cirrhosis are more commonly linked to late recurrence.^7–9 While transcatheter arterial chemoembolization (TACE) is commonly used in this setting, the introduction of immune checkpoint inhibitors (ICIs) offers a novel therapeutic option. Building on recent trials investigating ICIs in advanced or unresectable HCC, such as IMbrave150 and CheckMate 459, attention has shifted to their potential in earlier disease stages. However, limited data exist regarding their application following surgery, especially in patients with MVI. Moreover, the duration of ICI therapy necessary to maximize efficacy remains undefined. This study sought to compare ICI-based therapy with TACE in a real-world postoperative setting and to assess whether extended ICI treatment (≥12 months) offers added benefit in reducing recurrence.

Patients and Methods

Patients and Study Design

To evaluate the comparative efficacy of adjuvant ICI therapy versus TACE in improving RFS among HCC patients with MVI, we conducted a retrospective, multicenter cohort study enrolling patients who underwent curative-intent hepatic resection between January 1, 2017, and March 31, 2024, at Peking Union Medical College Hospital (Beijing), China-Japan Friendship Hospital (Beijing), and Sun Yat-sen University Cancer Center (Guangzhou). Inclusion criteria were: (1) complete (R0) resection; (2) age between 18 and 75 years; (3) pathologically confirmed HCC with MVI; and (4) Eastern Cooperative Oncology Group performance status (ECOG PS) score ≤ 1. Exclusion criteria encompassed: (1) recurrent HCC following prior curative resection; (2) history of spontaneous tumor rupture with hemorrhage; (3) coexistence of other active malignancies, including those in sustained remission; (4) administration of any neoadjuvant treatment modalities for HCC, including TACE, molecular targeted therapy, immunotherapy, or radiotherapy; and (5) death from non-HCC-related causes prior to follow-up.

Treatment and Data Collection

For patients anticipated to undergo extensive hepatectomy, preoperative evaluation of hepatic functional reserve was performed using the indocyanine green (ICG) retention test at 15 minutes. Surgical procedures were standardized across centers in accordance with established protocols previously described in the literature.¹⁰

Postoperative follow-up was routinely conducted 4 to 8 weeks after hepatic resection. For patients with MVI, adjuvant therapy was recommended contingent upon satisfactory general health and absence of contraindications. Given the lack of universally accepted adjuvant standards for MVI-positive HCC, treatment decisions were made based on multidisciplinary clinical evaluation, institutional practice patterns, and physician consensus. Regimen selection followed consistent clinical principles across centers and reflected contemporary real-world management of high-risk HCC. All adjuvant therapies, when applied, were initiated within 4 to 8 weeks following surgery.

Adjuvant TACE was initiated within 4–6 weeks post-resection, involving femoral artery catheterization and infusion of chemotherapeutic agents, as per institutional protocols.

Frequently employed targeted agents included lenvatinib,¹¹ apatinib,¹² donafenib, regorafenib,¹³ bevacizumab,¹⁴ administered once daily with dosage adjusted according to body weight or manufacturer guidelines. Immunotherapeutic agents commonly utilized in the adjuvant setting included sintilimab,¹⁵ carrelizumab,¹⁶ atezolizumab,¹⁴ tislelizumab.¹⁷ Prior to initiation, patients underwent thorough pre-treatment screening comprising blood panels, thyroid function assessment, electrocardiography, and chest CT imaging to ensure eligibility. ICIs were administered intravenously every three weeks in accordance with recommended dosing protocols. Vital signs were closely monitored during infusion and for at least 1hour post-infusion to promptly detect and manage any infusion-related reactions. Although the intended duration of adjuvant ICI therapy was 24 months, actual treatment durations varied in clinical practice. Some patients discontinued treatment within the first few months due to adverse events, financial burden, or poor adherence, even in the absence of tumor recurrence. To reduce immortal time bias in survival comparisons, patients who received ICI therapy for less than 6 months or experienced disease recurrence within 6 months were excluded from the duration-based survival analysis.

Follow-Up

Postoperative surveillance was rigorously implemented. All patients were scheduled for monthly follow-up visits during the first three months after hepatic resection, followed by assessments every three months for the subsequent two years, and semiannually thereafter. Each follow-up visits included evaluation of serum tumor markers, abdominal ultrasonography, and contrast-enhanced abdominal CT or MRI. Additional investigations, such as chest CT, bone scan, or positron emission tomography–CT (PET-CT), were performed when distant metastasis was clinically suspected. Follow-up was continued until patient death or loss to follow-up. The endpoint of the follow-up was September 20, 2024. Treatment of recurrence was personalized according to tumor profile, organ function, and patient status. The primary endpoint was RFS, as it reflects the direct effect of adjuvant therapy on preventing early relapse, particularly relevant in MVI-positive patients. OS was designated as a secondary endpoint, acknowledging the variability introduced by post-recurrence treatment heterogeneity. Of the 319 patients who received postoperative adjuvant therapy, 30 (9.4%) were lost to follow-up. Among the 80 patients under routine surveillance who did not receive adjuvant treatment, 9 (11.2%) were lost to follow-up. These individuals were not excluded from the analysis; instead, the time of their last documented follow-up was incorporated as censored observations in the survival analysis. This censoring strategy, implemented via the Kaplan–Meier method, ensured the inclusion of all available patient data and upheld the statistical robustness and validity of survival estimations.

Statistical Analysis

Continuous variables were summarized as mean ± standard deviation or median with interquartile range (IQR), and compared using Student’s t test or Mann–Whitney U-test, as appropriate. Categorical variables were compared using Pearson’s chi-square or Fisher’s exact test. To adjust for baseline differences between the TACE and immunotherapy groups, 1:1 PSM was performed using a nearest-neighbor algorithm (caliper = 0.2, no replacement). Covariates included AFP grade, HBV DNA, tumor differentiation, surgical margin, tumor number and size, satellite nodules, tumor embolus, MVI, and liver cirrhosis, selected based on clinical relevance and potential influence on treatment assignment. RFS and OS were estimated by Kaplan–Meier analysis and compared using Log rank tests; Cox regression identified independent predictors. Variables with P < 0.05 in univariate analysis, as well as clinically relevant covariates, were entered into the multivariate Cox regression model. To reduce immortal time bias, a 6-month landmark analysis excluded patients with recurrence or death before this point. Analyses were conducted using R (v4.3.1), with p < 0.05 considered significant.

Results

Patient Characteristics

From January 2017 to March 2024, a total of 1526 HCC patients from the three aforementioned centers were initially enrolled. All patients underwent curative liver resection, and postoperative pathology confirmed the diagnosis of HCC. Among them, 1,048 patients without MVI, 46 patients who received neoadjuvant therapy, 23 patients diagnosed with concurrent malignancies, and 10 patients who died from non-HCC-related causes were excluded. Ultimately, 399 patients were included in the final analysis (Figure 1). Among these patients, 132 received TACE alone, 58 received TACE combined with targeted therapy, 68 received TACE combined with targeted immunotherapy, 21 received targeted therapy combined with immunotherapy, 40 received only immunotherapy, and the remaining 80 patients did not receive any form of postoperative adjuvant therapy. Among the 319 patients who received postoperative adjuvant therapy, 30 were lost to follow-up, and among the 80 patients under active monitoring, 9 were lost to follow-up.

Figure 1 Patients flow chart. Abbreviations: MVI, microvascular invasion; HCC, hepatocellular carcinoma; TACE, transcatheter arterial chemoembolization; ICI, immune checkpoint inhibitor.

A total of 129 patients received adjuvant immunotherapy following curative hepatic resection. Among them, 40 received tislelizumab (anti–PD-1), 24 received camrelizumab (anti–PD-1), 21 received sintilimab (anti–PD-1), 11 received atezolizumab (anti–PD-L1), 10 received cadonilimab (a bispecific PD-1/CTLA-4 antibody), 10 received toripalimab (anti–PD-1), 7 received envafolimab (anti–PD-L1), and 6 received pembrolizumab (anti–PD-1). To minimize immortal time bias in the treatment duration analysis, 45 patients were excluded due to tumor recurrence occurring between 2 and 6 months postoperatively or receipt of immunotherapy for fewer than 6 months. The remaining 84 patients were included in the final analysis: 46 received adjuvant immunotherapy for less than 12 months, and 38 received it for 12 months or longer (Figure 1).

Before matching, the TACE and immunotherapy groups were largely comparable in baseline demographics and disease characteristics, except for differences in age, HBV DNA, and AFP levels. After 1:1 propensity score matching, 108 patients were included in each group (Figure 1), with no significant differences observed in baseline characteristics. A summary of baseline characteristics before and after matching is presented in Table 1.

Table 1 Baseline Characteristics of HCC Patients in the TACE and Immunotherapy Groups Before and After PSM

Treatment and Efficacy

During a median follow-up of 18 months (IQR 10–29 months), recurrence or metastasis occurred in 44 patients (34.1%) in the immunotherapy cohort and 86 patients (65.2%) in the TACE cohort. The predominant recurrence sites were the liver, lungs, and bones.

The median RFS was significantly longer in the immunotherapy cohort at 35 months (95% CI, 19–NA), compared with 16 months (95% CI, 10.8–27) in the TACE cohort (HR = 0.50, 95% CI, 0.34–0.72; p = 0.00015; Figure 2a). RFS rates at 12, 24, and 36 months were 75.4%, 53.1%, and 49.6% in the immunotherapy group, versus 54.5%, 42.8%, and 25.9% in the TACE group. Mortality was lower in the immunotherapy cohort (5.4%, 7 patients) compared to the TACE cohort (17.4%, 23 patients). Median overall survival (OS) was not reached in either group, but OS was significantly improved with immunotherapy (HR = 0.34, 95% CI, 0.14–0.80; p = 0.0096; Figure 2b). OS rates at 12, 24, and 36 months were 100.0%, 93.6%, and 86.9% in the immunotherapy cohort, versus 88.6%, 82.8%, and 81.0% in the TACE cohort. After PSM, the RFS benefit associated with immunotherapy remained statistically significant (HR = 0.54, 95% CI, 0.36–0.82; p = 0.0042; Figure 2c), while the OS difference was no longer statistically significant (HR = 0.43, 95% CI, 0.18–1.40; p = 0.060; Figure 2d).

Figure 2 Kaplan-Meier survival curves comparing the adjuvant ICI cohort and the TACE cohort: (a) RFS and (b) OS before PSM; (c) RFS and (d) OS after PSM. Abbreviations: HR, hazard ratio; CI, confidence interval.

Univariate and Multivariate Analyses of RFS and OS

Univariate and multivariate Cox regression analyses identified several independent predictors of poor prognosis (Table 2). For RFS, an advanced CNLC stage was significantly associated with shorter survival. For OS, a resection margin of less than 0.5 cm was identified as independent adverse prognostic factors.

Table 2 Univariate and Multivariate Analysis for RFS and OS of HCC Patients

Adverse Events (AE) in the Immunotherapy Cohort

Among the 129 patients who received immunotherapy, 55 patients (42.6%) experienced at least one treatment-related AE. Grade 1–2 AEs occurred in 51 patients (39.5%), while 13 patients (10.1%) experienced grade 3–4 events. No grade 5 AEs were reported (Supplementary Table 1). The most frequent AEs (any grade) were rash (9.3%), elevated AST (8.5%), ALT (7.8%), thrombocytopenia (7.0%), hypoalbuminemia (6.2%), and hypothyroidism (6.2%). Most events were grade 1–2 in severity. The most common grade 3–4 AEs included rash (5.4%), hypothyroidism (4.7%), and hypertension (3.9%). Other observed toxicities such as increased bilirubin (3.9%), elevated creatinine (3.9%), leukopenia (3.1%), mouth ulcers (3.1%), and fatigue (1.6%) were generally mild and manageable. No treatment-related deaths were observed.

Efficacy of Immunotherapy Duration on Survival Outcomes and AEs

To minimize immortal time bias, we excluded patients with recurrence-free survival less than 6 months and those who received ICI therapy for fewer than 6 months. Among the remaining cohort, patients who received adjuvant ICI therapy for 12 months or longer demonstrated significantly better RFS compared to those treated for less than 12 months (HR: 0.46, 95% CI: 0.21–0.99, p = 0.041; Figure 3a). A similar trend toward improved overall survival was observed, although the difference did not reach statistical significance (HR: 0.19, 95% CI: 0.02–1.59, p = 0.086; Figure 3b).

Figure 3 Survival outcomes of patients receiving adjuvant ICI therapy for ≥12 months versus <12 months: (a) RFS and (b) OS. Abbreviations: HR, hazard ratio; CI, confidence interval.

Among the 129 patients who received ICI-based adjuvant therapy, 68 received TACE combined with targeted immunotherapy, 21 received targeted immunotherapy, and 40 received immunotherapy alone. To account for the potential influence of different treatment regimens on survival outcomes, we conducted a subgroup analysis focusing on the largest group—patients who received postoperative TACE combined with targeted immunotherapy. After excluding those with a DFS less than 6 months and those who received immunotherapy for less than 6 months, we reevaluated RFS and OS. The analysis showed that patients who received ICIs for 12 months or longer had significantly improved RFS compared to those treated for less than 12 months (HR: 0.29, 95% CI: 0.11–0.79, p = 0.011; Figure 4a). While a longer OS was also observed in patients treated for 12 months or more, this difference was not statistically significant (HR: 0.19, 95% CI: 0.02–1.61, p = 0.089; Figure 4b).

Figure 4 Survival outcomes in the subgroup receiving TACE combined with targeted immunotherapy, stratified by ICI treatment duration (≥12 months vs <12 months): (a) RFS and (b) OS. Abbreviations: HR, hazard ratio; CI, confidence interval.

Among patients who received adjuvant immunotherapy, no significant differences were observed in the total number of adverse events between those treated for ≥12 months and those treated for <12 months (84.8% vs 89.5%, p >0.999; Supplementary Table 2). Similarly, the occurrence of grade 3–4 adverse events did not differ significantly between the two groups (26.1% vs 15.9%, p = 0.149; Supplementary Table 2). When comparing specific adverse events, no individual AE type showed a statistically significant difference between the two groups. However, numerically higher rates of hypertension and thrombocytopenia were noted in the ≥12-month treatment group, suggesting a trend toward increased incidence of some events with prolonged immunotherapy. Overall, extended treatment duration was not associated with a significantly increased risk of severe toxicity.

Discussion

Adjuvant therapies such as TACE, targeted agents, and immunotherapy have been associated with improved RFS and OS in HCC patients after curative resection.^18–20 However, the optimal duration of adjuvant immunotherapy for HCC patients with high-risk recurrence factors remains undefined, and real-world evidence on this topic is lacking. Although certain guidelines recommend that adjuvant immunotherapy should not exceed one year, they do not specify a minimum or preferred treatment duration.^21,22 In our study, compared to TACE alone, immunotherapy—either as monotherapy or in combination with TACE or targeted agents—was associated with a significant reduction in recurrence and improvement in OS among HCC patients with MVI who underwent R0 resection. After PSM, the benefit in RFS remained statistically significant, while the difference in overall survival was attenuated and no longer reached statistical significance. Importantly, ICI-based adjuvant therapies did not lead to a significant increase in AEs, with most AEs being grade 1–2, indicating good safety and tolerability. Among patients receiving adjuvant ICIs, a treatment duration of 12 months or longer was associated with significantly improved RFS compared to shorter durations. While a numerically favorable trend in OS was noted in the longer-duration group, the difference was not statistically significant. Therefore, no definitive conclusion regarding OS benefit can be drawn based on the current data, and this observation should be interpreted with caution. Furthermore, the total number of AEs and the incidence of grade 3–4 AEs were not significantly increased with longer treatment durations. Nevertheless, given the limited sample size of our study, larger and more comprehensive trials are needed to validate the safety and efficacy of postoperative adjuvant ICI therapy in this setting.

The clinical efficacy of ICIs was initially established in the advanced or unresectable HCC setting, as demonstrated by trials such as CheckMate 459²³ and IMbrave150.⁷ These studies showed that ICIs enhance antitumor immunity and may eradicate disseminated tumor cells. Extending this principle to earlier disease stages, adjuvant therapy aims to eliminate residual micrometastases and reduce distant recurrence, a major cause of treatment failure in MVI-positive patients. Indeed, several recent trials have tested this hypothesis. IMbrave050 reported an early RFS benefit with atezolizumab plus bevacizumab, though its updated analysis raised concerns about durability.²⁴ In contrast, a Phase II randomized controlled trial investigating adjuvant sintilimab showed a significant improvement in RFS among HCC patients with MVI,¹⁹ aligning with our findings and underscoring that high-risk populations may derive the greatest benefit. Retrospective studies have also suggested that adjuvant ICIs may improve prognosis among patients at high risk of recurrence.^25,26

Although several studies have investigated the safety and efficacy of adjuvant immunotherapy for HCC, the optimal duration of treatment has not been thoroughly explored. Given that treatment duration may critically influence patient outcomes, there is an urgent need for dedicated clinical trials addressing this issue. However, research specifically focused on treatment duration remains scarce. A prospective, multicenter cohort study evaluated the impact of adjuvant ICI treatment duration on RFS and OS in HCC patients at high risk of recurrence.²⁷ The results suggested that patients receiving adjuvant ICI therapy for more than six months tended to achieve better RFS and OS compared to those treated for six months or less, although the differences did not reach statistical significance. Despite the absence of a positive finding, the study indicated that six months of adjuvant ICI therapy might be insufficient and that extended treatment duration could potentially yield greater clinical benefits. Importantly, the design of ongoing Phase III randomized trials also reflects this rationale. Major studies such as CheckMate-9DX, KEYNOTE-937, JUPITER-04, SHR-1210-III-325, EMERALD-2, and DaDaLi have all adopted a 12-month adjuvant ICI regimen as the standard duration,²⁷ underscoring the clinical plausibility of our chosen cutoff. Nevertheless, the optimal duration of adjuvant immunotherapy remains an unresolved issue, not only in HCC but also in other malignancies such as non-small cell lung cancer^28,29 and melanoma, where prolonged ICI therapy has shown improved outcomes in certain settings. Our findings suggest that extending ICI therapy beyond 12 months may confer additional benefits for high-risk HCC patients; however, this hypothesis requires validation in prospective randomized studies. Future research should focus on defining the optimal duration of adjuvant immunotherapy, identifying predictive biomarkers for treatment benefit, and developing combination strategies tailored to individual recurrence risk profiles.

As a retrospective study, our analysis is inevitably subject to inherent biases. We acknowledge that comparisons based on treatment duration are vulnerable to immortal time bias, as longer-lived patients may be more likely to receive prolonged therapy. To mitigate this issue, we excluded patients who experienced recurrence or death within 6 months after surgery and those who received ICI therapy for less than 6 months. By restricting the analysis to patients who survived at least 6 months and initiated ICI treatment early, the impact of immortal time bias was reduced, although residual confounding remains possible. We also recognize that analyses involving secondary endpoints and subgroup comparisons may increase the risk of type I error. Another limitation is that the majority of ICIs used in our cohort were PD-1 inhibitors, with only a minority of patients treated with a bispecific PD-1/CTLA-4 antibody. Consequently, the efficacy and safety of other classes of immunotherapeutic agents, such as PD-L1 inhibitors, CTLA-4 inhibitors, and dual-targeting antibodies, were not assessed and warrant further investigation. Moreover, the heterogeneity of treatment regimens within our cohort may have influenced the outcomes. In addition, most patients in our study had HBV-related HCC, reflecting the epidemiological profile of HBV-endemic regions. Therefore, the generalizability of our findings to populations with HCV-related, alcohol-related, or non-viral HCC remains uncertain. Taken together, these limitations indicate that our conclusions should be interpreted with caution. Nonetheless, our findings provide important insights into the potential inadequacy of immunotherapy durations shorter than one year in high-risk HCC patients and highlight the need for prospective, standardized studies across diverse patient populations to confirm these observations.

Conclusions

This retrospective cohort study suggests that adjuvant ICI therapy following curative resection may improve RFS in HCC patients at high risk of recurrence compared to TACE. Notably, our findings indicate that a treatment duration of 12 months or longer is associated with improved RFS in patients with MVI. However, no statistically significant improvement in OS was observed with longer treatment duration. These results highlight the need to reconsider adjuvant immunotherapy strategies in this population, and underscore the importance of prospective, randomized, and large-scale clinical trials to determine the optimal duration of adjuvant ICI therapy for HCC.

Abbreviations

HCC, Hepatocellular carcinoma; MVI, Microvascular invasion; ICI, Immune checkpoint inhibitor; TACE, Transcatheter arterial chemoembolization; RFS, Recurrence-free survival; OS, Overall survival; PSM, Propensity score matching; AFP, Alpha-fetoprotein; HAIC, Hepatic arterial infusion chemotherapy; ECOG PS, Eastern Cooperative Oncology Group performance status; CT, Computed tomography; MRI, Magnetic resonance imaging; ICG, Indocyanine green; AST, Aspartate aminotransferase; ALT, Alanine aminotransferase; CNLC, China Liver Cancer staging system; BCLC, Barcelona Clinic Liver Cancer staging; PET-CT, Positron emission tomography-computed tomography; HR, Hazard ratio; CI, Confidence interval; AE, Adverse event; PD-1, Programmed death-1; PD-L1, Programmed death-ligand 1; CTLA-4, Cytotoxic T-lymphocyte-associated protein 4.

Data Sharing Statement

All data supporting the results of the study can be found in the article. Further inquiries can be directed to the corresponding author.

Statement of Ethics

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Peking Union Medical College Hospital (Approval No. I-23PJ964). Informed consent was obtained from all individual participants included in the study.

Acknowledgments

Xiaokun Chen, Jiali Xing, and Baoluhe Zhang are co-first authors for this study. We thank all the patients and the medical staff.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

This work was supported by the National Natural Science Foundation of China (81972698); the CAMS Innovation Fund for Medical Sciences (CIFMS 2021-I2M-01-014); Changzhou Xi Tai Hu development foundation for frontier cell- therapeutic technology (2024-P-019); the 2024 PhD Short-term Academic Visiting Program of Peking Union Medical College; the Start-up Fund from the Department of Liver Surgery, Peking Union Medical College Hospital; the Central high-level hospital clinical research special key cultivation project (2022-PUMCH-C-047); and 2021 Liver Cancer Diagnosis and Treatment Exchange Fund of Hubei Chen Xiaoping Science and Technology Development Foundation (CXPJJH1200009-01).

Disclosure

The authors have no conflicts of interest to declare in this work.

References

1. Han B, Zheng R, Zeng H, et al. Cancer incidence and mortality in China, 2022. J Natl Cancer Cent. 2024;4(1):47–53. doi:10.1016/j.jncc.2024.01.006

2. Rumgay H, Arnold M, Ferlay J, et al. Global burden of primary liver cancer in 2020 and predictions to 2040. J Hepatol. 2022;77(6):1598–1606. doi:10.1016/j.jhep.2022.08.021

3. Zhou L, Wang S-B, Chen S-G, et al. Risk factors of recurrence and poor survival in curatively resected hepatocellular carcinoma with microvascular invasion. Adv Clin Exp Med. 2020;29(7):887–892. doi:10.17219/acem/76750

4. Lee S, Kang TW, Song KD, et al. Effect of microvascular invasion risk on early recurrence of hepatocellular carcinoma after surgery and radiofrequency ablation. Ann Surg. 2021;273(3):564–571. doi:10.1097/SLA.0000000000003268

5. Straś WA, Wasiak D, Łągiewska B, et al. Recurrence of hepatocellular carcinoma after liver transplantation: risk factors and predictive models. Ann Transplant. 2022;27:e934924. doi:10.12659/AOT.934924

6. Hirokawa F, Hayashi M, Asakuma M, et al. Risk factors and patterns of early recurrence after curative hepatectomy for hepatocellular carcinoma. Surg Oncol. 2016;25(1):24–29. doi:10.1016/j.suronc.2015.12.002

7. Qin S, Chen M, Cheng A-L, et al. Atezolizumab plus bevacizumab versus active surveillance in patients with resected or ablated high-risk hepatocellular carcinoma (IMbrave050): a randomised, open-label, multicentre, Phase 3 trial. Lancet. 2023;402(10415):1835–1847. doi:10.1016/S0140-6736(23)01796-8

8. Pan H, Zhou L, Cheng Z, et al. Perioperative Tislelizumab plus intensity modulated radiotherapy in resectable hepatocellular carcinoma with macrovascular invasion: a phase II trial. Nat Commun. 2024;15(1):9350. doi:10.1038/s41467-024-53704-5

9. Peng N, Mao L-F, Su J-Y, et al. The efficacy and safety of tislelizumab with or without tyrosine kinase inhibitor as adjuvant therapy in hepatocellular carcinoma with high-risk of recurrence after curative resection. Front Immunol. 2025;16:1593153. doi:10.3389/fimmu.2025.1593153

10. Shi M, Guo R-P, Lin X-J, et al. Partial hepatectomy with wide versus narrow resection margin for solitary hepatocellular carcinoma: a prospective randomized trial. Ann Surg. 2007;245(1):36–43. doi:10.1097/01.sla.0000231758.07868.71

11. Kudo M, Finn RS, Qin S, et al. Lenvatinib versus sorafenib in first-line treatment of patients with unresectable hepatocellular carcinoma: a randomised phase 3 non-inferiority trial. Lancet. 2018;391(10126):1163–1173. doi:10.1016/S0140-6736(18)30207-1

12. Qin S, Li Q, Gu S, et al. Apatinib as second-line or later therapy in patients with advanced hepatocellular carcinoma (AHELP): a multicentre, double-blind, randomised, placebo-controlled, phase 3 trial. Lancet Gastroenterol Hepatol. 2021;6(7):559–568. doi:10.1016/S2468-1253(21)00109-6

13. Bruix J, Qin S, Merle P, et al. Regorafenib for patients with hepatocellular carcinoma who progressed on sorafenib treatment (RESORCE): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet. 2017;389(10064):56–66. doi:10.1016/S0140-6736(16)32453-9

14. Zhu AX, Abbas AR, de Galarreta MR, et al. Molecular correlates of clinical response and resistance to atezolizumab in combination with bevacizumab in advanced hepatocellular carcinoma. Nat Med. 2022;28(8):1599–1611. doi:10.1038/s41591-022-01868-2

15. Ren Z, Xu J, Bai Y, et al. Sintilimab plus a bevacizumab biosimilar (IBI305) versus sorafenib in unresectable hepatocellular carcinoma (ORIENT-32): a randomised, open-label, Phase 2–3 study. Lancet Oncol. 2021;22(7):977–990. doi:10.1016/S1470-2045(21)00252-7

16. Qin S, Chan SL, Gu S, et al. Camrelizumab plus rivoceranib versus sorafenib as first-line therapy for unresectable hepatocellular carcinoma (CARES-310): a randomised, open-label, international phase 3 study. Lancet. 2023;402(10408):1133–1146. doi:10.1016/S0140-6736(23)00961-3

17. Qin S, Kudo M, Meyer T, et al. Tislelizumab vs sorafenib as first-line treatment for unresectable hepatocellular carcinoma: a phase 3 randomized clinical trial. JAMA Oncol. 2023;9(12):1651–1659. doi:10.1001/jamaoncol.2023.4003

18. Wen Y, Lu L, Mei J, et al. Hepatic arterial infusion chemotherapy vs transcatheter arterial chemoembolization as adjuvant therapy following surgery for MVI-positive hepatocellular carcinoma: a multicenter propensity score matching analysis. J Hepatocell Carcinoma. 2024;11:665–678. doi:10.2147/JHC.S453250

19. Wang K, Xiang Y-J, Yu H-M, et al. Adjuvant sintilimab in resected high-risk hepatocellular carcinoma: a randomized, controlled, phase 2 trial. Nat Med. 2024;30(3):708–715. doi:10.1038/s41591-023-02786-7

20. Chen X, Wu X, Peng W, et al. Combined TACE with targeted and immunotherapy versus TACE alone improves DFS in HCC with MVI: a multicenter propensity score matching study. J Hepatocell Carcinoma. 2025;12:561–577. doi:10.2147/JHC.S504016

21. Benson AB, D’Angelica MI, Abrams T, et al. NCCN clinical practice guidelines in Oncology. (NCCN Guidelines^®). Hepatobiliary cancers. 2024 (Version 2.2024).

22. Singal AG, Llovet JM, Yarchoan M, et al. AASLD practice guidance on prevention, diagnosis, and treatment of hepatocellular carcinoma. Hepatology. 2023;78(6):1922–1965. doi:10.1097/HEP.0000000000000466

23. Yau T, Park J-W, Finn RS, et al. Nivolumab versus sorafenib in advanced hepatocellular carcinoma (CheckMate 459): a randomised, multicentre, open-label, phase 3 trial. Lancet Oncol. 2022;23(1):77–90. doi:10.1016/S1470-2045(21)00604-5

24. Yopp A, Kudo M, Chen M, et al. LBA39 Updated efficacy and safety data from IMbrave050: phase III study of adjuvant atezolizumab (atezo) + bevacizumab (bev) vs active surveillance in patients (pts) with resected or ablated high-risk hepatocellular carcinoma (HCC). Ann Oncol. 2024;35.

25. Chen W, Hu S, Liu Z, et al. Adjuvant anti-PD-1 antibody for hepatocellular carcinoma with high recurrence risks after hepatectomy. Hepatol Int. 2023;17(2):406–416. doi:10.1007/s12072-022-10478-6

26. Yang J, Jiang S, Chen Y, et al. Adjuvant ICIs plus targeted therapies reduce HCC recurrence after hepatectomy in patients with high risk of recurrence. Curr Oncol. 2023;30(2):1708–1719. doi:10.3390/curroncol30020132

27. Su J-Y, Liu SP, Xu XL, et al. Treatment duration of adjuvant immune checkpoint inhibitors in hepatocellular carcinoma patients at high risk of recurrence after resection: a prospective, multicentric cohort study. Liver Cancer. 2024;1–18.

28. Waterhouse DM, Garon EB, Chandler J, et al. Continuous versus 1-year fixed-duration nivolumab in previously treated advanced non-small-cell lung cancer: checkMate 153. J Clin Oncol. 2020;38(33):3863–3873. doi:10.1200/JCO.20.00131

29. Zalcman G, Balmaña J, Bober SL, et al. 972O Nivolumab (Nivo) plus ipilimumab (Ipi) 6-months treatment versus continuation in patients with advanced non-small cell lung cancer (aNSCLC): results of the randomized IFCT-1701 phase III trial. Ann Oncol;2022:33. doi:10.1016/j.annonc.2022.10.004

Introduction

Severe pneumonia is a major infectious disease in elderly patients, often characterized by rapid progression, multiple organ involvement, and poor prognosis.^1,2 One of its most severe complications is acute respiratory distress syndrome (ARDS), which arises from diffuse alveolar-capillary damage and manifests as refractory hypoxemia, respiratory distress, and decreased lung compliance.^3–5 Due to age-related immune decline, impaired pulmonary reserve, and high comorbidity burden, elderly patients are particularly vulnerable to developing ARDS after pneumonia, with mortality reaching 30–50%.^6,7 Despite advances in supportive care and precision medicine approaches, early diagnosis and prognosis assessment of ARDS still rely mainly on clinical criteria, imaging, and arterial blood gases,⁸ which lack sensitivity and specificity. This underscores the urgent need for reliable molecular biomarkers to improve early identification, risk stratification, and clinical decision-making.

MicroRNAs (miRNAs) are endogenous, non-coding small RNAs that regulate gene expression at the post-transcriptional level by binding to the 3′-UTR of target mRNAs. They are involved in processes such as inflammation, oxidative stress, apoptosis, and immune regulation.^9,10 miR-27a has been implicated in multiple inflammatory and malignant diseases, where it modulates immune signaling and cellular stress responses.^11,12 In pulmonary studies, miR-27a has been linked to anti-inflammatory and antioxidant effects, suggesting its potential role as a biomarker of lung injury and prognosis.¹³

Forkhead box O3 (FOXO3), a transcription factor downstream of the PI3K/Akt pathway, plays a central role in regulating apoptosis, oxidative stress responses, and inflammatory mediator release.^14,15 Previous reports indicate its involvement in lung tissue injury and ARDS pathogenesis.^16,17 Importantly, FOXO3 is a validated target of miR-27a: downregulation of miR-27a leads to FOXO3 activation, thereby promoting inflammatory cascades and exacerbating tissue damage.¹⁸

However, evidence regarding the specific expression patterns of miR-27a and FOXO3 in elderly patients with severe pneumonia complicated with ARDS remains scarce. In particular, their correlation with oxygenation index and short-term prognosis has not been fully elucidated. Therefore, this study retrospectively analyzed elderly patients with severe pneumonia, aiming to (1) detect serum levels of miR-27a and FOXO3, (2) assess their relationship with ARDS severity and 28-day mortality, and (3) evaluate their predictive value as biomarkers for prognosis. These findings may provide a molecular basis for improved survival assessment and risk stratification in elderly ARDS patients.

Materials and Methods

Study Subjects

This was a retrospective observational study including a total of 189 elderly inpatients (aged ≥60 years) with severe pneumonia admitted to the intensive care unit of our hospital from February 2023 to October 2024. According to whether ARDS was present, patients were divided into two groups: Group A (n=114, with ARDS) and Group B (n=75, without ARDS). Additionally, 70 healthy volunteers undergoing physical examination at the physical examination center during the same period were selected as the healthy control group. Among the 114 patients in Group A, there were 70 males and 44 females, with an average age of (73.16±6.72) years; in Group B, there were 46 males and 29 females, with an average age of (74.11±6.98) years; in the control group, there were 41 males and 29 females, with an average age of (73.24±7.19) years. There were no statistically significant differences in gender and age among the three groups (P>0.05), indicating comparability.

The sample size was estimated based on a preliminary analysis of 40 patients in our institution, where the difference in serum miR-27a expression between ARDS and non-ARDS patients was approximately 0.8 standard deviations. Using a two-sided α=0.05 and power (1–β)=0.80, the minimum required sample size per group was calculated as 64 cases. Considering potential dropouts and missing data, the final enrollment exceeded this requirement, ensuring adequate statistical power for group comparisons.

This study was approved by the Liberation Army General Hospital Medical Ethics Committee (Approval No.: 2024ZZLS12) and conducted in strict accordance with the ethical principles of the Declaration of Helsinki. All participants provided informed consent and signed relevant informed consent forms.

Inclusion and Exclusion Criteria

Inclusion criteria: (1) Age ≥60 years, regardless of gender; (2) First diagnosis of severe pneumonia or severe pneumonia with ARDS; (3) Diagnosis of severe pneumonia meets the criteria of the “Chinese Expert Consensus on Clinical Practice of Severe Pneumonia in Emergency Medicine”;¹⁹ (4) Diagnosis of ARDS conforms to the “Berlin Definition of Acute Respiratory Distress Syndrome”;²⁰ (5) Complete and reliable clinical data available for analysis.

Exclusion criteria: (1) Combined with malignant tumors, tuberculosis, HIV infection, or other immunodeficiency diseases; (2) Severe hepatic or renal failure, or heart failure; (3) Complicated with pulmonary tuberculosis, COPD, congenital lung dysplasia, or other pulmonary diseases; (4) History of immunosuppressant or hormone therapy within the past 6 months; (5) Complicated with severe infection at other sites or multiple organ failure; (6) Incomplete test data or improper specimen storage; (7) Pregnant or lactating women; (8) Considered unsuitable for inclusion by the researchers, such as those with psychiatric disorders or cognitive impairment who cannot cooperate with the study procedures.

Clinical Data

The following clinical data were collected for all enrolled patients: (1) Demographic data: gender; age; body mass index (BMI); smoking history; drinking history; living alone status; (2) Pneumonia-related information: type of pneumonia (community-acquired pneumonia/hospital-acquired pneumonia); whether complicated with underlying pulmonary diseases (eg, COPD, bronchiectasis, interstitial lung disease, etc).; (3) Underlying diseases: hypertension; diabetes; coronary heart disease; chronic liver disease; (4) Laboratory test indicators (within 24 h of admission): white blood cell count (WBC); C-reactive protein (CRP); procalcitonin (PCT); serum creatinine (Scr); blood urea nitrogen (BUN); (5) Mechanical ventilation: duration of mechanical ventilation.

All clinical data were extracted from the standardized hospital electronic medical record system, and only baseline information within 24 hours of admission was included to minimize bias. Data collection was independently performed by two trained researchers in a blinded manner, and any inconsistent entries were rechecked by a third investigator to ensure accuracy and reliability.

Grouping Method

Patients in Group A (with ARDS) were further subdivided according to the oxygenation index [arterial oxygen partial pressure (PaO₂) / fraction of inspired oxygen (FiO₂)] after admission: Mild ARDS subgroup: PaO₂/FiO₂ between 200–300 mmHg (n=28); Moderate ARDS subgroup: PaO₂/FiO₂ between 100–200 mmHg (n=36); Severe ARDS subgroup: PaO₂/FiO₂ <100 mmHg (n=50). Meanwhile, Group A patients were also stratified by their 28-day outcome into: Survival subgroup: patients who survived within 28 days (n=79); Death subgroup: patients who died within 28 days (n=35).

Detection of Serum miR-27a and FOXO3 mRNA

All enrolled patients underwent collection of a fasting early-morning venous blood sample (5 mL) within 24 hours of admission. Blood was placed in anticoagulant-free centrifuge tubes, left to clot at room temperature for 30 minutes, and then centrifuged at 3000 rpm for 10 minutes to isolate serum. The serum was immediately aliquoted into RNase-free centrifuge tubes (free of RNA contamination) and stored at –80°C in an ultra-low temperature freezer to ensure the quality and stability for subsequent RNA extraction. To detect the expression levels of miR-27a and FOXO3 mRNA in serum, qRT-PCR (quantitative real-time polymerase chain reaction) was used. The detection process included three main steps: total RNA extraction, reverse transcription, and real-time PCR amplification, detailed as follows: (1) Total RNA extraction: According to the instructions of the RNA extraction kit produced by Nanjing Vazyme Biotech Co., Ltd. (Product No.: RC112-01), total RNA was extracted from frozen serum. Strict RNase-free procedures were followed throughout. The concentration and purity of RNA samples were assessed using a NanoDrop™ UV spectrophotometer, and samples with A260/A280 between 1.8 and 2.1 were considered qualified. (2) Reverse transcription: RNA samples that passed quality assessment were reverse-transcribed into cDNA using the reverse transcription kit provided by Nanjing Saihongrui Biotech Co., Ltd. (Product No.: DV807A). Specific stem-loop primers were used for miRNA reverse transcription with U6 small nuclear RNA as the internal control, while mRNA reverse transcription was performed using a mixed system of Oligo(dT) and random primers, with GAPDH as the reference gene.(3) Real-time quantitative PCR amplification: The PCR reaction system was constructed based on the kit provided by Yeasen Biotechnology (Shanghai) Co., Ltd. (Product No.: 11203ES03). The total volume per reaction was 20 μL, including: 2× Master Mix buffer 10 μL, forward primer 0.5 μL, reverse primer 0.5 μL, cDNA template 2.0 μL, and RNase-free DEPC-treated water to 20 μL. Amplification was performed on an ABI 7500 real-time PCR system. Reaction conditions were: Pre-denaturation: 95°C for 30 seconds, 1 cycle; Amplification: 95°C for 5 seconds and 60°C for 40 seconds, 40 cycles; Melting curve: fluorescence signals were acquired from 65°C to 95°C, with 0.5°C increments to verify amplification specificity. (4) Data analysis and quality control: Fluorescence signal data were collected and calculated using the instrument’s built-in software. All raw amplification plots and melting curves were manually reviewed to confirm specificity. Samples with ambiguous amplification results were repeated to ensure reproducibility. Relative expression levels of miR-27a and FOXO3 mRNA were calculated using the 2^–ΔΔCt method, with U6 and GAPDH as internal controls for normalization, respectively. All primers were designed based on published sequences, verified in the NCBI database, and synthesized with quality certification by Wuhan GeneCreate Bioengineering Co., Ltd., and the primer sequences are shown in Table 1.

Table 1 Primer Sequence Information

Statistical Analysis

Statistical analysis was performed using SPSS 26.0 software, and figures were generated using GraphPad Prism 9.0. Continuous variables conforming to normal distribution were expressed as (); comparisons between two groups were made using the t-test, and comparisons among multiple groups were performed using analysis of variance (ANOVA). Categorical data were expressed as counts (n) and percentages (%), and comparisons were made using the χ²-test. Correlation analysis was performed using Pearson or Spearman correlation methods. Multivariate analysis used a binary logistic regression model to identify independent risk factors for 28-day mortality. ROC curves were plotted to compare the predictive performance of miR-27a, FOXO3, and their combination for patient mortality. Differences in AUC were compared using the Z-test. The significance level was set at α=0.05, with P<0.05 considered statistically significant.

Results

Comparison of Serum miR-27a and FOXO3 mRNA Levels Among the Three Groups

Serum miR-27a levels were significantly higher in Group B than in Group A, and further elevated in the control group compared with Group B. Conversely, FOXO3 mRNA levels were significantly lower in Group B than in Group A, and further reduced in the control group (F=77.352, 62.956, P<0.001), as shown in Figure 1. This indicates that compared with the control group, serum miR-27a levels were downregulated while FOXO3 mRNA levels were upregulated in the disease groups.

Figure 1 Comparison of serum miR-27a and FOXO3 mRNA levels among the three groups. Notes: *P<0.05 vs Group A; #P<0.05 vs Group B. All comparisons are made with reference to the control group.

Comparison of Serum miR-27a and FOXO3 mRNA Levels in Patients with Different Severities of Severe Pneumonia with ARDS

In patients with severe pneumonia and ARDS, serum miR-27a levels decreased progressively from the mild to moderate and severe subgroups, whereas FOXO3 mRNA levels increased in the same order (F=83.597, 111.834, P<0.001), as shown in Figure 2. These subgroups all belong to the disease group and reflect different severity classifications.

Figure 2 Comparison of serum miR-27a and FOXO3 mRNA levels in patients with different severities of severe pneumonia with ARDS. Notes: aP<0.05 vs mild subgroup; bP<0.05 vs moderate subgroup. All comparisons are relative within the disease group.

Correlation Between Serum miR-27a and FOXO3 mRNA Levels

Pearson correlation analysis showed that serum miR-27a and FOXO3 mRNA levels in elderly patients with severe pneumonia and ARDS were negatively correlated (r=–0.624, P<0.001), as shown in Figure 3.

Figure 3 Scatter plot of correlation between serum miR-27a and FOXO3 mRNA levels.

Spearman correlation analysis showed that the oxygenation index (mild=3, moderate=2, severe=1) was positively correlated with serum miR-27a levels (r=0.635, P<0.001), and negatively correlated with FOXO3 mRNA levels (r=–0.672, P<0.001), as shown in Figure 4. The oxygenation index was expressed in mmHg to ensure clarity.

Figure 4 Scatter plot of correlation between serum miR-27a, FOXO3 mRNA levels and oxygenation index (mmHg).

Comparison of Clinical Data in Patients with Different Prognoses of Severe Pneumonia with ARDS

The 28-day mortality rate in elderly patients with severe pneumonia and ARDS was 30.70% (35/114). The death subgroup had higher age, CRP, mechanical ventilation time, and FOXO3 mRNA levels, and lower oxygenation index and miR-27a levels compared to the survival subgroup (P<0.05). There were no statistically significant differences in other data (P>0.05), as shown in Table 2.

Table 2 Comparison of Clinical Data in Patients with Different Prognoses of Severe Pneumonia with ARDS

Multivariate Logistic Regression Analysis of Prognostic Factors in Elderly Patients with Severe Pneumonia and ARDS

Taking prognosis (survival=0, death=1) as the dependent variable, possible influencing factors from Table 1 were assigned as independent variables (see Table 3). A multivariate logistic regression model was established. Results showed that increased age, prolonged mechanical ventilation time, and elevated FOXO3 mRNA were independent risk factors, while increased oxygenation index and miR-27a levels were independent protective factors, as shown in Table 4.

Table 3 Variable Assignment Table

Table 4 Multivariate Logistic Regression Analysis of Prognostic Factors in Elderly Patients with Severe Pneumonia and ARDS

Predictive Value of miR-27a, FOXO3 mRNA, and Their Combination for Mortality in Elderly Patients with Severe Pneumonia and ARDS

The AUCs for serum miR-27a, FOXO3 mRNA, and their combination in predicting mortality in elderly patients with severe pneumonia and ARDS were 0.775, 0.781, and 0.867, respectively. The combined AUC was superior to each single index (Z_combined–miR-27a=2.557, P<0.05; Z_combined–FOXO3 mRNA=2.974, P<0.05), as shown in Table 5 and Figure 5.

Table 5 Predictive Value of miR-27a, FOXO3 mRNA, and Their Combination for Mortality in Elderly Patients with Severe Pneumonia and ARDS

Figure 5 ROC curves for predictive value of miR-27a, FOXO3 mRNA, and their combination in elderly patients with severe pneumonia and ARDS.

Discussion

This study focused on elderly patients with severe pneumonia complicated by ARDS, systematically analyzing the relationship between serum miR-27a and FOXO3 mRNA expression levels and disease severity and prognosis. Compared with healthy controls and patients with severe pneumonia without ARDS, ARDS patients showed significantly decreased serum miR-27a levels and markedly increased FOXO3 mRNA levels. Moreover, across different oxygenation index strata, miR-27a levels progressively declined with worsening ARDS severity, whereas FOXO3 mRNA levels increased stepwise relative to less severe subgroups. These findings indicate that these alterations are evident not only when compared with non-ARDS populations but also dynamically vary with disease progression, suggesting that both markers are closely associated with the pathophysiology of ARDS.

The observed inverse correlation between miR-27a and FOXO3 mRNA highlights a potential regulatory axis, in which miR-27a may play a protective role while FOXO3 promotes tissue damage. Previous studies^21,22 have confirmed that miR-27a regulates inflammatory, apoptotic, and oxidative stress pathways and plays crucial roles in various pulmonary diseases, including asthma, pulmonary fibrosis, and infections. Mechanistically, miR-27a may inhibit the release of pro-inflammatory mediators, limit oxidative damage, and reduce apoptosis by modulating NF-κB, TGF-β, and PI3K/Akt signaling pathways.^23–25 Downregulation of miR-27a weakens these protective effects, thereby amplifying inflammatory cascades. Conversely, persistent activation of FOXO3 exacerbates oxidative stress, induces mitochondrial dysfunction, and promotes immune imbalance.^26,27

Our results are consistent with the study by Lv et al,²⁸ who reported that downregulation of miR-27a aggravated alveolar injury in a murine ARDS model, whereas miR-27a mimic intervention effectively alleviated inflammation and tissue damage. FOXO3, on the other hand, is recognized as a transcription factor that promotes oxidative stress responses and cellular senescence. Wu et al²⁹ demonstrated that inhibition of FOXO3 could reduce alveolar epithelial apoptosis and preserve lung function. Together, these studies support the hypothesis that an imbalance between miR-27a and FOXO3 signaling contributes to the pathogenesis and progression of ARDS. Recent evidence also indicates that FOXO3 can influence macrophage polarization and T-cell differentiation, leading to immune dysregulation and impaired tissue repair.^30,31 These processes collectively create a vicious cycle of lung injury and inadequate repair, which aligns with the clinical features of refractory hypoxemia in elderly ARDS patients.

Clinically, our study found that the 28-day mortality rate among elderly ARDS patients reached 30.70%, higher than that reported for general ARDS populations,³² reflecting age-related vulnerability and the influence of comorbidities. Notably, multivariate logistic regression analysis indicated that elevated FOXO3 levels were an independent risk factor, whereas miR-27a and oxygenation index served as independent protective factors. ROC curve analysis showed that combined detection of these two markers achieved an AUC of 0.867, outperforming individual markers and providing a practical approach for risk stratification. Zhao et al³³ similarly demonstrated that multi-marker combined detection significantly improves prognostic prediction in ARDS patients. Therefore, this study expands the ARDS biomarker panel in elderly patients and validates the clinical utility of miR-27a and FOXO3.

In terms of novelty, this study has three main contributions. First, it is the first to combine the detection of miR-27a and FOXO3 mRNA in elderly ARDS patients, integrating molecular mechanisms with clinical prognostic assessment. Second, the inclusion of a relatively large cohort with stratification across ARDS severity enhances the clinical representativeness and reliability of the findings. Third, by focusing on elderly patients—a subgroup with poor outcomes that is often underrepresented in biomarker studies—this work fills a critical gap in ARDS research. These findings not only enrich current understanding but also provide a foundation for future therapeutic strategies targeting the miR-27a/FOXO3 signaling pathway.

However, several limitations should be acknowledged. First, as a single-center retrospective study, selection bias cannot be excluded, and multicenter prospective cohort studies are needed for validation. Second, only serum levels were assessed, lacking mechanistic validation in bronchoalveolar lavage fluid, lung tissue, or animal models. Moreover, miR-27a may regulate multiple targets beyond FOXO3, and FOXO3 may be influenced by other miRNAs or upstream signals; therefore, causal relationships remain to be confirmed. Functional experiments and multi-omics approaches could provide deeper insights into these interactions.

In conclusion, this study demonstrates that downregulated serum miR-27a and upregulated FOXO3 mRNA are closely associated with ARDS severity and short-term prognosis in elderly patients with severe pneumonia. Combined detection of these markers enhances predictive accuracy, providing a novel molecular basis for early identification, risk assessment, and potential therapeutic intervention. Future studies should integrate mechanistic validation and dynamic longitudinal monitoring to establish causal roles and explore their feasibility as intervention targets, ultimately advancing personalized management of ARDS.

Conclusion

The results of this study indicate that, compared with healthy controls and elderly patients with severe pneumonia without ARDS, serum miR-27a levels are significantly decreased, whereas FOXO3 mRNA levels are significantly increased in elderly patients with severe pneumonia complicated by ARDS. Within ARDS subgroups stratified by oxygenation index, miR-27a levels progressively decreased from mild to moderate to severe ARDS, while FOXO3 mRNA levels increased stepwise, highlighting their close association with disease severity. Elevated miR-27a may act as a protective factor, whereas elevated FOXO3 mRNA serves as an independent risk factor for poor short-term outcomes. Combined detection of these two biomarkers provides higher predictive efficacy for 28-day mortality than either marker alone, underscoring their potential utility for early risk stratification and clinical intervention. These findings suggest that the imbalance between miR-27a and FOXO3 is not only involved in the pathogenesis of ARDS but also has practical implications as prognostic biomarkers. Future studies should further investigate the molecular mechanisms underlying miR-27a regulation of FOXO3 and related downstream signaling pathways and validate their clinical utility in elderly ARDS populations through multicenter, large-sample prospective studies.

Funding

Project of National Clinical Research Center for Geriatric Diseases, Research on Diagnosis, Treatment and Comprehensive Prevention and Control Measures of Multidrug Resistant Pathogenic Bacteria Infections in Elderly Patients (Project Number: NCRCG-PLAGH-DX-2024002) 2. Military Health Care Project: Research on the Clinical Application Value of the Combined Evaluation Method of Pepsin, amylase and Lipid Cells in Tracheal Aspirates for the Diagnosis of Airway Aspiration (Project Number: 21BJZ24).

Disclosure

The authors report no conflicts of interest in this work.

References

1. Xu Y, Wang XY. [Clinical characters and influencing factors of perioperative pneumonia in elderly patients with hip fractures aged ≥80 years at a tertiary hospital in Beijing City]. Zhonghua Yu Fang Yi Xue Za Zhi. 2025;59(6):916–924. doi:10.3760/cma.j.cn112150-20250330-00257 Wolof

2. Kang M, Li J, Wan Q, et al. [Factors influencing the choice of endotracheal intubation and mechanical ventilation in patients with acute respiratory distress syndrome caused by viral pneumonia]. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2022;34(6):586–591. doi:10.3760/cma.j.cn121430-20220607-00549 Dutch

3. Zhang J, Li Y, Li H, et al. [Acute respiratory distress syndrome caused by severe respiratory infectious diseases: clinical significance and solution of maintaining artificial airway closure]. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2025;37(3):221–224. doi:10.3760/cma.j.cn121430-20240506-00404 Dutch

4. Tang R, Tang W, Wang D. [Predictive value of machine learning for in-hospital mortality for trauma-induced acute respiratory distress syndrome patients: an analysis using the data from MIMIC III]. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2022;34(3):260–264. doi:10.3760/cma.j.cn121430-20211117-01741 Dutch

5. Liu YN, Ma XC. [The precision medicine for the treatment of acute respiratory distress syndrome is based on identification of phenotypes]. Zhonghua Yi Xue Za Zhi. 2024;104(15):1253–1257. doi:10.3760/cma.j.cn112137-20230928-00603 Danish

6. Wu J, Xiao H, Li X, et al. [Evaluation value of sequential organ failure assessment score for predicting the prognosis of patients with acute respiratory distress syndrome due to severe pneumonia]. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2021;33(9):1057–1062. doi:10.3760/cma.j.cn121430-20210115-00076 Dutch

7. Zhang HL, Shang Y. [The value of point-of-care ultrasound in the diagnosis and management of acute respiratory distress syndrome]. Zhonghua Yi Xue Za Zhi. 2024;104(15):1225–1229. doi:10.3760/cma.j.cn112137-20230906-00407 Danish

8. Chalmers S, Khawaja A, Wieruszewski PM, et al. Diagnosis and treatment of acute pulmonary inflammation in critically ill patients: the role of inflammatory biomarkers. World J Crit Care Med. 2019;8(5):59–71. doi:10.5492/wjccm.v8.i5.74

9. Tian F, Ying H, Liao S, et al. lncRNA SNHG14 promotes the proliferation, migration, and invasion of thyroid tumour cells by regulating miR-93-5p. Zygote. 2022;30(2):183–193. doi:10.1017/S0967199421000319

10. Xing J, Chen LP, Yu WJ. [Non-small cell lung cancer-derived exosomal circular RNA circEZH2 activates fibroblasts by regulating the miR-495-3p / TPD52 axis and NF-κB pathway]. Zhonghua Zhong Liu Za Zhi. 2024;46(12):1176–1186. doi:10.3760/cma.j.cn112152-20230717-00015 Polish

11. Zhang G, Liu R, Dang X, et al. [Experimental study on improvement of osteonecrosis of femoral head with exosomes derived from miR-27a-overexpressing vascular endothelial cells]. Zhongguo Xiu Fu Chong Jian Wai Ke Za Zhi. 2021;35(3):356–365. doi:10.7507/1002-1892.202011026 Danish

12. Xie E, Lin M, Sun Z, et al. Serum miR-27a is a biomarker for the prognosis of non-small cell lung cancer patients receiving chemotherapy. Transl Cancer Res. 2021;10(7):3458–3469. doi:10.21037/tcr-20-3276

13. Fan X, Wang J, Qin T, et al. Exosome miR-27a-3p secreted from adipocytes targets ICOS to promote antitumor immunity in lung adenocarcinoma. Thorac Cancer. 2020;11(6):1453–1464. doi:10.1111/1759-7714.13411

14. Ruan P, Zheng Y, Dong Z, et al. [Research progress in the regulation of autophagy and mitochondrial homeostasis by AMPK signaling channels]. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2024;36(4):425–429. doi:10.3760/cma.j.cn121430-20230302-00132 Dutch

15. Yang F, Zhang H, Zhuo M, et al. [Effects of Peiminine on biological behavior and chemotherapy resistance of ovarian cancer by regulating the FOXO3-FOXM1 signaling axis]. Xi Bao Yu Fen Zi Mian Yi Xue Za Zhi. 2024;40(11):976–982. Wolof

16. Tao Y, Li -L-L, Liu S-H, et al. [FOXO3a signaling pathway in prostate cancer: progress in studies]. Zhonghua Nan Ke Xue. 2020;26(8):745–750. Basque

17. Wang Y, Liang H-X, Zhang C-M, et al. FOXO3/TRIM22 axis abated the antitumor effect of gemcitabine in non-small cell lung cancer via autophagy induction. Transl Cancer Res. 2020;9(2):937–948. doi:10.21037/tcr.2019.12.33

18. Zhang H, Xia J, Hu Q, et al. Long non‑coding RNA XIST promotes cerebral ischemia/reperfusion injury by modulating miR‑27a‑3p/FOXO3 signaling. Mol Med Rep. 2021;24(2):566.

19. Shang WF, Chen DC. [Prone positioning ventilation therapy in acute respiratory distress syndrome: knowns and unknowns in clinical efficacy]. Zhonghua Yi Xue Za Zhi. 2024;104(15):1236–1241. doi:10.3760/cma.j.cn112137-20231012-00720 Danish

20. Yuan XY, Liu L, Qiu HB. [New 2023 global definition of acute respiratory distress syndrome: progress and limitation]. Zhonghua Yi Xue Za Zhi. 2024;104(15):1216–1220. doi:10.3760/cma.j.cn112137-20231016-00770 Danish

21. Dong M, Wang X, Li T, et al. miR-27a-3p alleviates lung transplantation-induced bronchiolitis obliterans syndrome (BOS) via suppressing Smad-mediated myofibroblast differentiation and TLR4-induced dendritic cells maturation. Transpl Immunol. 2023;78:101806. doi:10.1016/j.trim.2023.101806

22. Sun Y, Jiang R, Hu X, et al. CircGSAP alleviates pulmonary microvascular endothelial cells dysfunction in pulmonary hypertension via regulating miR-27a-3p/BMPR2 axis. Respir Res. 2022;23(1):322. doi:10.1186/s12931-022-02248-7

23. Abd-Elhakim YM, Mohamed AA-R, Khamis T, et al. Alleviative effects of green-fabricated zinc oxide nanoparticles on acrylamide-induced oxidative and inflammatory reactions in the rat stomach via modulating gastric neuroactive substances and the MiR-27a-5p/ROS/NF-κB axis. Tissue Cell. 2024;91:102574. doi:10.1016/j.tice.2024.102574

24. Ong J, Faiz A, Timens W, et al. Marked TGF-β-regulated miRNA expression changes in both COPD and control lung fibroblasts. Sci Rep. 2019;9(1):18214. doi:10.1038/s41598-019-54728-4

25. Zhao XR, Zhang Z, Gao M, et al. MicroRNA-27a-3p aggravates renal ischemia/reperfusion injury by promoting oxidative stress via targeting growth factor receptor-bound protein 2. Pharmacol Res. 2020;155:104718. doi:10.1016/j.phrs.2020.104718

26. Shen J, Wang H, Wang J-S, et al. [The relationship between MicroRNA expression profiling in imatinib-resistant cell line K562/G and potential mechanism through FOXO3/Bcl-6 signaling pathway]. Zhongguo Shi Yan Xue Ye Xue Za Zhi. 2022;30(1):107–112. doi:10.19746/j.cnki.issn.1009-2137.2022.01.017 Hausa

27. Artham S, Gao F, Verma A, et al. Endothelial stromelysin1 regulation by the forkhead box-O transcription factors is crucial in the exudative phase of acute lung injury. Pharmacol Res. 2019;141:249–263. doi:10.1016/j.phrs.2019.01.006

28. Lv X, Zhang X-Y, Zhang Q, et al. lncRNA NEAT1 aggravates sepsis-induced lung injury by regulating the miR-27a/PTEN axis. Lab Invest. 2021;101(10):1371–1381. doi:10.1038/s41374-021-00620-7

29. Wu Z, Wang Y, Lu S, et al. SIRT3 alleviates sepsis-induced acute lung injury by inhibiting pyroptosis via regulating the deacetylation of FoxO3a. Pulm Pharmacol Ther. 2023;82:102244. doi:10.1016/j.pupt.2023.102244

30. Duan XH, Li H, Lyu Y, et al. Regulation of baicalin on growth of extranodal NK/T cell lymphoma cells through FOXO3/CCL22 signaling pathway. Zhongguo Shi Yan Xue Ye Xue Za Zhi. 2023;31(3):730–738. doi:10.19746/j.cnki.issn.1009-2137.2023.03.017

31. Wang H, Wang Z. Oroxylin-A attenuates taurocholate-induced lung injury via NF-κB pathway suppression. Clin Mol Epidemiol. 2025;2:5. doi:10.53964/cme.2025005

32. Wang HF, Hu WH, Song QW, et al. [Clinical study on the relationship between the exosomes in bronchoalveolar lavage fluid and plasma and the severity of lung injury and outcome in early acute respiratory distress syndrome patients]. Zhonghua Yi Xue Za Zhi. 2022;102(13):935–941. doi:10.3760/cma.j.cn112137-20211105-02448 Danish

33. Zhao C, Li Y, Wang Q, et al. Establishment of risk prediction nomograph model for sepsis related acute respiratory distress syndrome. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2023;35(7):714–718. doi:10.3760/cma.j.cn121430-20230215-00088