Factors associated with cancer-related fatigue in advanced hepatocellular carcinoma patients during the peri-intervention period: a path analysis model Authors: Sha Lin, Shanzhi Gu, Ling Li, Zhen Liang, Juan Li, Yulian Yuan Published: 2026 ABSTRACT Cancer-related fatigue (CRF) is a significant concern in oncology, yet limited research has focused on its associated factors and pathways in hepatocellular carcinoma (HCC). This study aimed to identify the factors associated with CRF in advanced HCC patients and to explore the interrelationships among these factors using pathway analysis. This cross-sectional study enrolled participants from a tertiary cancer hospital. Data were collected through self-administered questionnaires assessing demographics, disease characteristics, positive emotions, physical activity (PA), and CRF. Multiple linear regression analyses and pathway modeling were employed to examine the associations among these variables. A total of 220 out of 275 eligible participants were included in the study. The mean Cancer Fatigue Scale (CFS) score was 31.30 (SD = 6.160, 95% CI: 30.20–32.40), with a clinical fatigue prevalence of 94.5%. Multiple linear regression analysis revealed that several factors were significantly associated with CRF, including the Herth Hope Index (HHI) (β=-0.281, 95% CI: 0.75–1.34, P < 0.01), insomnia (β = 0.22, 95% CI: 0.84–1.19, P < 0.01), economic burden (β=-0.20, 95% CI: 0.94–1.06, P < 0.01), Child-Pugh score (β = 0.19, 95% CI: 0.86–1.06, P < 0.01), General Self-Efficacy Scale (GSES) (β=-0.18, 95% CI: 0.81–1.24, P < 0.01), body mass index (BMI) (β=-0.13, 95% CI: 0.83–1.20, P < 0.01), marital status (β=-0.12, 95% CI: 0.95–1.06, P < 0.01), and Prognostic Nutritional Index (PNI) (β = 0.12, 95% CI: 0.92–1.09, P < 0.01). The pathway model showed HHI had the greatest correlation with CRF followed by GSES, Child-Pugh, and ISI. PA-Index indirectly related CRF mainly through HHI and GSES. This study highlights the high prevalence and severity of CRF in advanced HCC patients. Interventions aimed at improving physical activity, managing liver function, fostering positive emotions, enhancing sleep quality, and optimizing nutritional status may effectively reduce the risk of CRF in this population. Introduction Hepatocellular carcinoma (HCC) was the sixth most commonly diagnosed cancer and the third leading cause of cancer-related deaths worldwide in 2020, with approximately 906,000 new cases and 830,000 deaths1. HCC is characterized by high incidence, high mortality, and poor prognosis, and it imposes a significant symptom burden on patients2,3. Most patients are diagnosed at middle or advanced stages, and interventional therapy is the primary treatment modality. The combination of interventional therapy with systemic treatments, including radiotherapy, chemotherapy, targeted therapy, and immunotherapy, represents a key focus and future direction in the development of interventional approaches4. Among the symptoms experienced by HCC patients, cancer-related fatigue (CRF) is one of the most severe5. According to the National Comprehensive Cancer Network (NCCN), CRF is defined as “a multifactorial syndrome characterized by a distressing, persistent sense of physical, emotional, and/or cognitive tiredness or exhaustion related to cancer or cancer treatment”6. The prevalence of CRF is as high as 85%, with 9% to 45% of patients reporting moderate-to-severe fatigue, and 21% to 52% of survivors continuing to experience CRF for up to 3 years post-diagnosis7–9. Despite its high prevalence, the pathogenesis of CRF remains poorly understood. CRF significantly impairs patients’ quality of life10,11 and is associated with increased mortality rates among cancer patients12. Therefore, identifying the risk factors associated with CRF is crucial for developing effective and targeted interventions to prevent or mitigate its occurrence and progression. Previous studies have identified several factors associated with CRF, including pain, insomnia, psychological symptoms, perceived social support, and physical activity (PA)13–16. Sociodemographic factors, such as age, sex, marital status15,17,18, monthly income, and education level16,19, have also been reported to influence CRF. Additionally, disease-related factors, such as time since diagnosis, symptom distress17,18, type of disease20, clinical stage, comorbidities, and treatment modalities15,21,22, play significant roles. In a real-world study23, hepatic arterial infusion chemotherapy combined with anti-PD-1/PD-L1 immunotherapy and molecularly targeted agents for advanced HCC revealed that CRF was one of the most common adverse events. Furthermore, psychological factors, such as hope level and self-efficacy, have been identified as key protective factors against CRF24–26. Several studies have also highlighted the close relationship between physical activity (PA) and CRF, noting that both higher body mass index (BMI) (often associated with obesity) and lower BMI (indicating malnutrition) may indirectly exacerbate CRF by reducing physical activity levels17,27. Despite extensive research on CRF and its influencing factors, most studies have focused on cancers with lower tumor invasiveness and symptom burden, such as breast cancer and prostate cancer28,29. In contrast, CRF has received less attention in more aggressive cancers with shorter survival times, such as pancreatic cancer and advanced liver cancer. On the one hand, poor liver function and the use of combination therapies (e.g., interventional therapy, targeted therapy, and immunotherapy) in advanced HCC patients often lead to occult hepatic encephalopathy, which manifests as physical and cognitive fatigue30. Consequently, CRF is particularly pronounced in advanced liver cancer. On the other hand, traditional Chinese medicine posits that long-term liver deficiency leads to qi and blood stasis in liver cancer patients, resulting in CRF that persists longer than in other cancer types31. Moreover, to date, few studies have attempted to examine how these factors collectively contribute to fatigue or to explore potential interactions among them. The theory of unpleasant symptoms32 suggests that symptoms are influenced by three main factors: physiological, psychological, and social/physical factors. Based on this theory, combined with findings from previous studies and the unique characteristics of HCC, this study hypothesizes that PA, insomnia, liver function, and psychological factors directly affect CRF, and that PA may also indirectly influence CRF through its effects on insomnia and psychological factors. Therefore, the purpose of this study was to identify the factors associated with CRF in advanced HCC patients during the peri- intervention period and to explore the interrelationships among these factors using path analysis, guided by the theory of unpleasant symptoms. Prevalence and severity of CRF were also described to contextualize the burden of fatigue in this population. Methods This research is a cross-sectional study that strictly complies with the ethical standards outlined in the Declaration of Helsinki and adheres to the World Medical Association‘s ethical principles. This study adheres to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement for reporting cross-sectional studies. The Ethics Committee for Clinical Trials at Hunan Cancer Hospital approved the study protocol (Approval No. SBQLL-2021-174, date: 25 August 2021). Participants HCC patients were recruited using a cross-sectional survey design and purposive sampling from the Department of Tumor Intervention at a tertiary cancer hospital in China, during the period from April to December 2022. Qualified research nurses selected and enrolled patients according to the inclusion and exclusion criteria. Each participant was provided with an information leaflet explaining the study’s objectives and procedures and gave written informed consent. The inclusion criteria were as follows: (1) a pathological confirmation of HCC diagnosis; (2) currently receiving or planning to receive interventional therapy as advised by the medical team; (3) within the peri‑intervention period, defined as the time from 7 days before the interventional procedure to 7 days after the procedure (including pre‑treatment preparation, the day of intervention, and early post‑procedure recovery); (4) the ability to communicate orally; and (5) age of 18 years or older. Exclusion criteria included: (1) a previous history of other malignant tumors or significant complications, such as severe heart or lung diseases, active malignant tumors, or systemic infections; and (2) a history of psychiatric illnesses. Sample size Based on the Kendall sample size estimation method for cross-sectional correlation studies33, the required sample size was calculated as 10–20 times the number of independent variables. We included 20 variables. Therefore, the minimum sample size was 20 × 10 = 200. Accounting for a potential 20% invalid response rate, the final target sample size was 200 ÷ (1 − 0.20) = 250. A total of 275 eligible patients were approached, and 220 completed the survey, exceeding the required minimum. Patient-reported assessments We approached and invited the eligible participants to participate in this study after admission. All participants were informed about purpose, benefit, or harms of the study. The research assistant trained in data collection, gave paper questionnaires at the hospital, assisted the patients if they had difficulty answering the survey by themselves. After patients completed the questionnaire, the investigator checked whether items were completed correctly and ensured that no missing data were collected. Measures 2.4.1 Cancer Fatigue Scale The Cancer Fatigue Scale (CFS) was developed by Okuyama et al.34 in 2000 to evaluate CRF symptoms in cancer patients, and was later translated and revised by Zhang35.The scale contains 15 items subscales: physical (7 items with total score of 28), emotional (4 items with total score of 16), and cognitive (4 items with total score of 16). Each item is scored on a 5-point Likert scale from “nothing” to “very much”, score of 0–4 and score of 0–60, higher scores indicate fatigue. The cutoff value of CFS is 18/19 and the total score of CFS > 18 points is “clinical fatigue”. Cronbach’s alpha coefficients for this scale were 0.87 (physical), 0.81 (affective), 0.76 (cognitive), and 0.84 (total). Cronbach’s alpha coefficient for this sample was 0.759. Physical activity Physical activity (PA) was assessed with the Leisure Score Index from the Godin–Leisure Time Exercise Questionnaire (GLTEQ), a tool commonly used in studies of cancer patients36. The GLTEQ asks participants to report the frequency, during the previous week, of three intensities of leisure-time activity: vigorous (e.g., activities that markedly increase heart rate such as running, swimming, basketball), moderate (e.g., activities causing a noticeable heart-rate increase such as brisk walking, square dancing), and light (e.g., activities that maintain a normal heart rate such as casual walking, yoga, Tai Chi). Participants self-reported their exercise during the week before hospitalization. The PA index representing total weekly activity was calculated as follows: (9 × frequency of vigorous exercise/week) + (5 × frequency of moderate exercise/week) + (3 × frequency of light exercise/week). Herth Hope Index This study used the Chinese version of the Herth Hope Index (HHI) revised by Zhao37, originally developed by Herth38. The questionnaire contains 12 items organized into three dimensions: positive attitudes toward reality and the future (4 items), positive actions (4 items), and close relationships with others (4 items). Each item is rated on a four-point scale from 1 to 4, corresponding to “strongly disagree,” “disagree,” “agree,” and “strongly agree.” Higher total scores indicate greater levels of hope. Score ranges classify hope as low (12–23), moderate (24–35), or high (36–48). The overall Cronbach’s coefficient for the scale is 0.969, and 0.842 for this sample. General Self‑Efficacy The General Self‑Efficacy Scale (GSES) was developed by Schwarzer et al.39 and translated and revised by Wang et al.40. The GSES is a one-dimensional scale with 10 items scored on a 4-point Likert scale: 1 (completely wrong), 2 (basically right), 3 (almost right), 4 (absolutely right). A higher score indicates better general self-efficacy. The scale is satisfactory internal consistency (Cronbach a = 0.863). Insomnia Severity Index The Insomnia Severity Index (ISI)41 assesses insomnia symptoms using seven items scored on a 0–28 scale. The Chinese version of this scale can reliably and sensitively reflect patients’ clinical changes42. Scores of 0–7 indicate no clinically significant insomnia, 8–14 indicate mild insomnia, 15–21 indicate moderate insomnia, and 22–28 indicate severe insomnia. The Cronbach’s alpha for this sample is 0.767. Sociodemographic, clinical variables, health-related and behavioral factors Sociodemographic characteristics comprised age, gender, education, marital status, social support, and economic stress. Disease-related variables included disease duration, stage, treatment, comorbidity, liver function score, pain, and hemoglobin. Health factors comprised BMI and PNI, where PNI = serum albumin (g/L) + 5 × lymphocyte count (109/L); PNI was dichotomized as ≤ 45 (low) versus > 45 (positive). Social support was measured as perceived availability of help from friends and family on a 0–10 scale (0 = not available; 10 = available when needed). Behavioral factors included the PA-index. Statistical methods Data analysis was performed using IBM SPSS version 26.0 for windows. Descriptive statistics were used to summarize sample characteristics and outcome variables. Continuous outcome variables were shown as mean and standard deviation (SD). Counting data were described using frequencies and percentages; Independent samples t-test or ANOVA was used for one-way analysis; Pearson or Spearman’s was used for correlation analysis; and multiple linear regression analysis was used for multifactor analysis. Path analysis was performed using AMOS 24.0. Differences were considered statistically significant at P (two-tailed) < 0.05. Results First, we reached 275 eligible colorectal cancer patients, 35 refused, 12 didn’t fill the questionnaire half the time and 8 checked the same option in all questionnaires, 220 were added, effective response rate of 80%. The average CFS score was 31.30 (SD = 6.160, range 6–57 (95% CI: 30.20–32.40), prevalence rate of 94.5% (208/220) for clinical fatigue (CFS score 18). The three dimensions of CFS mean scores physical fatigue was 12.95 (SD = 4.78, range 2–27 (95% CI: 12.31–13.58); emotion fatigue was 8.31(SD = 2.51, range 0–14 (95% CI: 7.97–8.64); cognitive fatigue was 10.06 (SD = 2.76, range 0–20 (95% CI: 9.70–10.43). Table 1 presents the composition of the demographic and clinical characteristics of the sample, as well as comparisons with the CRF states. Approximately 82.3% of the sample was below 65 years old and 79.1% were male. The majority had an education level of middle school or below (67.3%), approximately 10% of the sample had no spouse (Unmarried, widowed, divorced), and approximately 74% had a heavy financial burden. The majority was primary liver cancer (75%), approximately79% of the sample had a liver function score of Child-A, and 37.7% experienced pain at the time of the survey at the time of the survey. Approximately 64% of patients had disease duration of less than one year, and 71.8% had stage C cancers. The differences in CRF scores for gender, education, marriage, economic stress, cancer type (primary or secondary), liver function stage, tumor stage, pain, treatment modality, and hemoglobin level were not statistically significant in this study (all P < 0.05). Table 1Demographic and clinical characteristics, description of CRF score(N = 220).CharacteristicsFrequencyN(%)CRFMean (SD)t/F p Age (years) ≤ 5064 (29.1)30.05 (8.87)0.880.672 51–65117 (53.2)32.05 (8.33) ≥ 6539 (17.7)31.10 (7.19)Gender Male174 (79.1)30.41 (8.30)3.180.002 Female46 (20.9)34.70 (7.54)Education level ≤Primary school37 (16.8)34.11 (7.20)1.500.040 Junior high school111 (50.5)30.41 (7.54) College or high school52 (23.6)31.92 (9.29) ≥University20 (9.1)29.40 (10.64)Marital status Without spouse20 (10)27.71 (8.28)2.070.039 Married200 (90)31.67 (8.25)Economic burden Very serious97 (44.1)33.86 (7.84)1.900.003 Heavier66 (30.0)31.38 (7.97) Commonly46 (20.9)27.41 (7.45) Lighter11 (5.0)24.55 (8.99)Cancer type Primary liver cancer167 (75.9)30.70 (8.18)2.030.044 Metastatic liver cancer53 (24.1)33.31 (8.491)Liver function grading (child–pugh) Child-A165 (79.7)29.68 (7.13)3.280.000 Child-B42 (20.3)34.84 (8.91) Child-C3 (1.4)54.00 (8.32)Pain 0129 (62.3)29.72 (8.36)2.160.000 1–260 (29.0)32.75 (6.12) ≥ 318 (8.7)37.23 (9.98)Time from diagnosis (years) < 1142 (64.5)30.84 (8.28)0.990.500 1–244 (20.0)31.41 (8.16) 2–316 (7.3)34.56 (7.47) > 318 (8.2)31.78 (9.62)Cancer staging (BCLC) B57 (25.9)29.32 (8.34)2.250.026 C158 (71.8)32.18 (8.15) Missing3 (2.3)21.33 (9.24)Treatment Untreated52 (23.6)29.87 (8.71)0.780.024 Interventional therapy51 (23.2)30.63 (8.26) Targeted/immunotherapy20 (9.1)30.65 (5.13) Interventional +targeted/immunotherapy97 (44.1)32.56 (8.58)Comorbidities No133 (60.5)31.50 (8.75)0.480.63 Yes87 (39.5)30.97 (7.64)Anemia (HB) No (≥ 120 g/L)181 (82.3)30.41 (7.78)1.580.024 Mild (90–119 g/L)268 (12.7)35.68 (9.65) Moderate (60–89 g/L)9 (4.1)34.44 (10.41) Heavy (< 60 g/L)1 (0.5)30.00“Without spouse” refers to unmarried, divorced, or widowed individuals.Comorbidities include high blood pressure, diabetes, heart disease, and respiratory diseases.Child-Pugh is a clinical grading system that quantifies liver reserve function in patients with liver cancer; Grades A, B, and C indicate increasing severity of liver impairment (a higher score denotes worse reserve function).BCLC refers to the Barcelona Clinic Liver Cancer staging system, which classifies disease as A (early), B (intermediate), or C (advanced). Table 1Demographic and clinical characteristics, description of CRF score(N = 220).CharacteristicsFrequencyN(%)CRFMean (SD)t/F p Age (years) ≤ 5064 (29.1)30.05 (8.87)0.880.672 51–65117 (53.2)32.05 (8.33) ≥ 6539 (17.7)31.10 (7.19)Gender Male174 (79.1)30.41 (8.30)3.180.002 Female46 (20.9)34.70 (7.54)Education level ≤Primary school37 (16.8)34.11 (7.20)1.500.040 Junior high school111 (50.5)30.41 (7.54) College or high school52 (23.6)31.92 (9.29) ≥University20 (9.1)29.40 (10.64)Marital status Without spouse20 (10)27.71 (8.28)2.070.039 Married200 (90)31.67 (8.25)Economic burden Very serious97 (44.1)33.86 (7.84)1.900.003 Heavier66 (30.0)31.38 (7.97) Commonly46 (20.9)27.41 (7.45) Lighter11 (5.0)24.55 (8.99)Cancer type Primary liver cancer167 (75.9)30.70 (8.18)2.030.044 Metastatic liver cancer53 (24.1)33.31 (8.491)Liver function grading (child–pugh) Child-A165 (79.7)29.68 (7.13)3.280.000 Child-B42 (20.3)34.84 (8.91) Child-C3 (1.4)54.00 (8.32)Pain 0129 (62.3)29.72 (8.36)2.160.000 1–260 (29.0)32.75 (6.12) ≥ 318 (8.7)37.23 (9.98)Time from diagnosis (years) < 1142 (64.5)30.84 (8.28)0.990.500 1–244 (20.0)31.41 (8.16) 2–316 (7.3)34.56 (7.47) > 318 (8.2)31.78 (9.62)Cancer staging (BCLC) B57 (25.9)29.32 (8.34)2.250.026 C158 (71.8)32.18 (8.15) Missing3 (2.3)21.33 (9.24)Treatment Untreated52 (23.6)29.87 (8.71)0.780.024 Interventional therapy51 (23.2)30.63 (8.26) Targeted/immunotherapy20 (9.1)30.65 (5.13) Interventional +targeted/immunotherapy97 (44.1)32.56 (8.58)Comorbidities No133 (60.5)31.50 (8.75)0.480.63 Yes87 (39.5)30.97 (7.64)Anemia (HB) No (≥ 120 g/L)181 (82.3)30.41 (7.78)1.580.024 Mild (90–119 g/L)268 (12.7)35.68 (9.65) Moderate (60–89 g/L)9 (4.1)34.44 (10.41) Heavy (< 60 g/L)1 (0.5)30.00“Without spouse” refers to unmarried, divorced, or widowed individuals.Comorbidities include high blood pressure, diabetes, heart disease, and respiratory diseases.Child-Pugh is a clinical grading system that quantifies liver reserve function in patients with liver cancer; Grades A, B, and C indicate increasing severity of liver impairment (a higher score denotes worse reserve function).BCLC refers to the Barcelona Clinic Liver Cancer staging system, which classifies disease as A (early), B (intermediate), or C (advanced). Demographic and clinical characteristics, description of CRF score(N = 220). “Without spouse” refers to unmarried, divorced, or widowed individuals. Comorbidities include high blood pressure, diabetes, heart disease, and respiratory diseases. Child-Pugh is a clinical grading system that quantifies liver reserve function in patients with liver cancer; Grades A, B, and C indicate increasing severity of liver impairment (a higher score denotes worse reserve function). BCLC refers to the Barcelona Clinic Liver Cancer staging system, which classifies disease as A (early), B (intermediate), or C (advanced). Table 2 shows the linear correlation between CRF and some continuous variables. CRF was significantly positively correlated with ISI and PNI (r = 0.42, r = 0.17, respectively; all p < 0.001). Similarly, CRF was significantly negatively correlated with PSS, PA-Index, HHI, GSES and BMI (r = 0.20, r = 0.42, r = 0.48, r = 0.39, r = 0.33, respectively; all p < 0.001). Table 2Correlation analysis of CRF, PSS, exercise, HHI, GSES, ISI, BMI, and PNI (r -value, n = 220).variablesCRFEmotional fatigueEmotional fatigueCognitive fatiguePSSPA-IndexHHIGSESISIBMIPNICRF1.000––––––––––Physical fatigue.6022)1.000–––––––––Emotional fatigue.8932).3722)1.000––––––––Cognitive fatigue.8092).3502).6122)1.000–––––––PSS− .1962)− .2442)− 0.113− .1792)1.000––––––PA-Index− .4152)− .3102)− .3802)− .3252)0.1321.000–––––HHI− .4802)− .3062)− .4632)− .4082)0.074.3942)1.000––––GSES− .3882)− .2152)− .3742)− .3562)0.082.3042).2962)1.000–––ISI.4162).2262).4212).3202)− 0.125− .1762)− .1752)− .3022)1.000––BMI− .3312)− .2072)− .3942)− .1882).1721).1762).2432).1591)− .2922)1.000–PNI.1661).1631).1631).1812)− .1361)0.038− .1571)− .1782)0.101− 0.0231.000(1) P<0.01; (2) P<0.05.CRF, Cancer-Related Fatigue; PSS, Perceived Social Support; HHI, Herth Hope Index; GSES, General Self-Efficacy; ISI, Insomnia Severity Index; BMI, Body-Mass Index; PA-Index, physical activity Index; PNI, Prognostic Nutrition Index. Table 2Correlation analysis of CRF, PSS, exercise, HHI, GSES, ISI, BMI, and PNI (r -value, n = 220).variablesCRFEmotional fatigueEmotional fatigueCognitive fatiguePSSPA-IndexHHIGSESISIBMIPNICRF1.000––––––––––Physical fatigue.6022)1.000–––––––––Emotional fatigue.8932).3722)1.000––––––––Cognitive fatigue.8092).3502).6122)1.000–––––––PSS− .1962)− .2442)− 0.113− .1792)1.000––––––PA-Index− .4152)− .3102)− .3802)− .3252)0.1321.000–––––HHI− .4802)− .3062)− .4632)− .4082)0.074.3942)1.000––––GSES− .3882)− .2152)− .3742)− .3562)0.082.3042).2962)1.000–––ISI.4162).2262).4212).3202)− 0.125− .1762)− .1752)− .3022)1.000––BMI− .3312)− .2072)− .3942)− .1882).1721).1762).2432).1591)− .2922)1.000–PNI.1661).1631).1631).1812)− .1361)0.038− .1571)− .1782)0.101− 0.0231.000(1) P<0.01; (2) P<0.05.CRF, Cancer-Related Fatigue; PSS, Perceived Social Support; HHI, Herth Hope Index; GSES, General Self-Efficacy; ISI, Insomnia Severity Index; BMI, Body-Mass Index; PA-Index, physical activity Index; PNI, Prognostic Nutrition Index. Correlation analysis of CRF, PSS, exercise, HHI, GSES, ISI, BMI, and PNI (r -value, n = 220). (1) P<0.01; (2) P<0.05. CRF, Cancer-Related Fatigue; PSS, Perceived Social Support; HHI, Herth Hope Index; GSES, General Self-Efficacy; ISI, Insomnia Severity Index; BMI, Body-Mass Index; PA-Index, physical activity Index; PNI, Prognostic Nutrition Index. In Table 3, Multiple linear regression analyses were performed on variables that were statistically significantly correlated with CRF, based on the findings of this study. For CRF, the variables HHI (β=-0.281, 95% CI: 0.75–1.34, P < 0.01), ISI (β = 0.22, 95% CI:0.84–1.19, P < 0.01), Economic burden (β=-0.20, 95% CI:0.94–1.06, P < 0.01), Child-Pugh (β = 0.19, 95% CI:0.86–1.16, P < 0.01), GSES (β=-0.18, 95% CI:0.81–1.24, P < 0.01), BMI (β=-0.13, 95% CI:0.83–1.20, P < 0.01), Marital status (β=-0.12, 95% CI:0.95–1.06, P < 0.01) and PNI (β = 0.12, 95% CI:0.92–1.09, P < 0.01) were statistically significant predictors. This model explained 53.7% for CRF in advanced liver cancer patients (R2 = 0.537). In this study, Amos23.0 was used to establish the corresponding path model; the initial path model was established based on the results of related analysis and existing literature reports as shown in Fig. 1. The model was fitted using the great likelihood method and corrected according to the correction path provided by the Modification Indices (MI) for the poor fitting of the initial model. The modified model is shown in Fig. 2. All the fitting indices were within the acceptable range (Chi-square / DF = 1.457, GFI = 0.964, AGFI = 0.932, TLI = 0.970, CFI = 0.980, IFI = 0.981, RMSEA = 0.046). The model shows that the PA-Index affects CRF mainly indirectly through HHI and GSES; the liver function staging can affect the CRF both directly and indirectly by affecting the PA-Index and HHI. The insomnia index can affect the CRF directly and at the same time it is affected by GSES. Table 3Multiple linear regression analysis of CRF and variables with statistically significant correlations.BSEβt P 95.0%CI for BCovariance statisticsLowerUpperToleranceVIF(Constant)55.7715.6179.9280.00055.7715.617HHI− 0.4640.088− 0.281−5.2490.0000.7471.338− 0.4640.088ISI0.2710.0640.2154.2720.0000.8431.1860.2710.064Economic burden−1.7730.433− 0.195−4.0910.0000.9431.060−1.7730.433Child-Pugh3.2940.8690.1893.7920.0000.8621.1603.2940.869GSES− 0.2300.067− 0.176−3.4240.0010.8071.240− 0.2300.067BMI− 0.3500.133− 0.134−2.6300.0090.8311.204− 0.3500.133Marital status−3.4651.367− 0.121−2.5350.0120.9481.055−3.4651.367PNI0.1220.0500.1172.4320.0160.9211.0850.1220.050R = 0.745, R2= 0.554, Adjusted R2= 0.537.Expression of value: Economic burden (1 = Very serious, 2 = serious,3 = normal, 4 = lighter); Child-Pugh (1 = Child-A, 2 = Child-B, 3 = Child-C); Marital status (1 = Married,2 = without spouse, Unmarried/divorced/widowed); HHI, ISI, GSES, BMI and PNI (Original value input).CRF, Cancer-Related Fatigue; HHI, Herth Hope Index; ISI, Insomnia Severity Index; GSES, General Self-Efficacy; BMI, Body-Mass Index (kg/m2); PNI, prognostic nutrition Index; Child-Pugh, a clinical grading system that quantifies liver reserve function in patients with liver cancer. Table 3Multiple linear regression analysis of CRF and variables with statistically significant correlations.BSEβt P 95.0%CI for BCovariance statisticsLowerUpperToleranceVIF(Constant)55.7715.6179.9280.00055.7715.617HHI− 0.4640.088− 0.281−5.2490.0000.7471.338− 0.4640.088ISI0.2710.0640.2154.2720.0000.8431.1860.2710.064Economic burden−1.7730.433− 0.195−4.0910.0000.9431.060−1.7730.433Child-Pugh3.2940.8690.1893.7920.0000.8621.1603.2940.869GSES− 0.2300.067− 0.176−3.4240.0010.8071.240− 0.2300.067BMI− 0.3500.133− 0.134−2.6300.0090.8311.204− 0.3500.133Marital status−3.4651.367− 0.121−2.5350.0120.9481.055−3.4651.367PNI0.1220.0500.1172.4320.0160.9211.0850.1220.050R = 0.745, R2= 0.554, Adjusted R2= 0.537.Expression of value: Economic burden (1 = Very serious, 2 = serious,3 = normal, 4 = lighter); Child-Pugh (1 = Child-A, 2 = Child-B, 3 = Child-C); Marital status (1 = Married,2 = without spouse, Unmarried/divorced/widowed); HHI, ISI, GSES, BMI and PNI (Original value input).CRF, Cancer-Related Fatigue; HHI, Herth Hope Index; ISI, Insomnia Severity Index; GSES, General Self-Efficacy; BMI, Body-Mass Index (kg/m2); PNI, prognostic nutrition Index; Child-Pugh, a clinical grading system that quantifies liver reserve function in patients with liver cancer. Multiple linear regression analysis of CRF and variables with statistically significant correlations. R = 0.745, R2= 0.554, Adjusted R2= 0.537. Expression of value: Economic burden (1 = Very serious, 2 = serious,3 = normal, 4 = lighter); Child-Pugh (1 = Child-A, 2 = Child-B, 3 = Child-C); Marital status (1 = Married,2 = without spouse, Unmarried/divorced/widowed); HHI, ISI, GSES, BMI and PNI (Original value input). CRF, Cancer-Related Fatigue; HHI, Herth Hope Index; ISI, Insomnia Severity Index; GSES, General Self-Efficacy; BMI, Body-Mass Index (kg/m2); PNI, prognostic nutrition Index; Child-Pugh, a clinical grading system that quantifies liver reserve function in patients with liver cancer. Fig. 1The initial path-model. CRF, Cancer-related fatigue; HHI, Herth Hope Index; ISI, The Insomnia Severity Index; PA-Index, physical activity index; GSES, General self-efficacy; Child-Pugh, the higher the score, the worse the liver reserve function. The initial path-model. CRF, Cancer-related fatigue; HHI, Herth Hope Index; ISI, The Insomnia Severity Index; PA-Index, physical activity index; GSES, General self-efficacy; Child-Pugh, the higher the score, the worse the liver reserve function. Fig. 2The final model with standardized correlation coefficients. CRF, Cancer-related fatigue; HHI, Herth Hope Index; ISI, The Insomnia Severity Index; PA-Index, physical Activity Index; GSES, General Self-Efficacy; Child-Pugh, the higher the score, the worse the liver reserve function. The final model with standardized correlation coefficients. CRF, Cancer-related fatigue; HHI, Herth Hope Index; ISI, The Insomnia Severity Index; PA-Index, physical Activity Index; GSES, General Self-Efficacy; Child-Pugh, the higher the score, the worse the liver reserve function. Table 4 shows the direct effect, indirect effect and total effect of each variable. In terms of total effect, HHI had the greatest influence on the CRF of advanced liver cancer patients, followed by GSES, Child–Pugh, ISI, and PA-Index. Table 4Direct, indirect, and total effects on CRF.Direct effectsIndirect effectsTotal effectsChild-Pugh0.220.150.37PA-Index−0.22−0.22GSES–0.21−0.19−0.40ISI0.320.32HHI−0.45−0.45Child-Pugh, the higher the score, the worse the liver reserve function.CRF, Cancer-related fatigue; HHI, Herth Hope Index; ISI, The Insomnia Severity Index; GSES, General self-efficacy; PA-Index, physical activity index. Table 4Direct, indirect, and total effects on CRF.Direct effectsIndirect effectsTotal effectsChild-Pugh0.220.150.37PA-Index−0.22−0.22GSES–0.21−0.19−0.40ISI0.320.32HHI−0.45−0.45Child-Pugh, the higher the score, the worse the liver reserve function.CRF, Cancer-related fatigue; HHI, Herth Hope Index; ISI, The Insomnia Severity Index; GSES, General self-efficacy; PA-Index, physical activity index. Direct, indirect, and total effects on CRF. Child-Pugh, the higher the score, the worse the liver reserve function. CRF, Cancer-related fatigue; HHI, Herth Hope Index; ISI, The Insomnia Severity Index; GSES, General self-efficacy; PA-Index, physical activity index. Discussion To our knowledge, this is the first study to comprehensively analyze the factors influencing CRF in advanced HCC patients during the peri-interventional period, incorporating PA, disease characteristics, psychological factors, and social factors. By constructing a pathway model, we elucidated the interrelationships and the direct and mediating effects of these factors on CRF. The findings of this study indicate that CRF is highly prevalent among advanced HCC patients during the peri-interventional period, with a mean CFS score of 31.3 (indicating severe fatigue) and a clinical fatigue incidence of 94.5%. Among the three CFS subscales, physical fatigue scored the highest, followed by cognitive fatigue, suggesting that CRF predominantly affects physical and cognitive functioning. The incidence and mean scores of CRF in this sample were higher than those reported in studies of colorectal and breast cancers, where the incidence of clinical fatigue was 69.4% and 43%, respectively17,21. Similarly, most studies have reported the highest scores in the physical fatigue dimension19,21,43. These discrepancies may be attributed to differences in disease type, clinical condition, disease duration, treatment modalities, and study design. In conclusion, given the high prevalence and severity of peri-interventional CRF in HCC patients, there is a critical need for healthcare providers to enhance the assessment and management of CRF. This pooled analysis aimed to identify associations between CRF and PA, psychological factors, insomnia, and physical condition (including pain, BMI, and PNI) in advanced HCC survivors. Given that patients with advanced HCC are at risk for occult hepatic encephalopathy—manifested by fatigue and cognitive impairment—due to combination therapies and poor liver function30, we hypothesized that both treatment modalities and liver function contribute to CRF. Among self-reported data, the Herth HHI, ISI, GSES, and economic burden were significantly correlated with CRF. Regarding objective data, BMI, PNI, liver function, and marital status were also associated with CRF. Contrary to expectations, PA, pain, and treatment modalities, which showed significant correlations in univariate analyses, were not statistically significant in the multivariate regression models. This discrepancy may be explained by several factors: First, the statistical power of the study may have been limited by the sample size, particularly given that only 8.7% of participants (n = 18) reported pain scores of 3 or higher, while the majority experienced no or mild pain that did not interfere with sleep. Second, the study population consisted of advanced HCC patients who already carried a significant psychological and symptomatic burden prior to treatment, suggesting that psychological factors may play a more prominent role in CRF than treatment modalities. Supporting this, other studies have found that cancer type and treatment regimen may not be as influential as previously assumed21. However, further longitudinal studies are needed to confirm these findings. It is noteworthy that PA was not associated with CRF after adjusting for covariates in the regression model, despite previous observational studies identifying inactivity as a significant correlate of CRF and PA as a key strategy for alleviating fatigue44,45. Interestingly, a study on breast cancer survivors similarly found no association between objectively measured PA and CRF17. To explore this inconsistency, we identified that PA was positively associated with HHI and GSES, which fully mediated the pathway from PA to CRF, with effect sizes of 0.45 and 0.21, respectively. While previous research has primarily focused on negative emotions such as anxiety and depression as contributors to CRF13,14,16, our findings highlight the protective role of positive emotions. For instance, optimism has been shown to have a direct negative effect on CRF and is a key pathway influencing fatigue46. Additionally, studies have demonstrated that even suboptimal levels of PA can reduce depressive symptoms by enhancing self-efficacy47,48. Therefore, interventions that promote hope and self-efficacy through PA may offer a promising approach to reducing CRF. In the pathway analysis, liver function and ISI were found to have a direct positive effect on CRF, with worse liver function and increased insomnia associated with more severe fatigue. These findings align with previous studies identifying insomnia as a significant contributor to CRF14,15. Interestingly, while a systematic review suggested that PA improves sleep disorders49, a prospective cohort study of 440 breast cancer patients aged over 65 found that increased PA was associated with reduced fatigue but had no effect on sleep disturbances50. Notably, our path analysis revealed that self-efficacy fully mediated the relationship between PA and insomnia. In other words, patients who engaged in PA experienced improved sleep quality by enhancing their confidence and ability to cope with their illness. Furthermore, liver function emerged as a critical variable influencing CRF in HCC patients. Poor liver function not only directly increased CRF (total effect = 0.37) but also negatively impacted hope levels and PA. This is likely due to the severe gastrointestinal symptoms, fatigue, and complications (e.g., ascites, electrolyte imbalances, and infections) associated with advanced liver disease, which can diminish patients’ confidence and motivation to engage in PA. Therefore, healthcare providers should prioritize liver function management in HCC patients, particularly when severe fatigue is observed post-treatment, as it may indicate impaired liver function. Higher fatigue levels were significantly associated with greater economic burden and poorer physical condition. Supporting evidence from other studies suggests that financial stress and healthcare costs are major contributors to CRF in HCC patients14,16. Given that low-income patients and those with limited insurance coverage often face significant financial pressures due to the need for multiple interventional treatments, it is crucial to address these burdens through targeted healthcare programs. Additionally, physical condition, as measured by BMI, PNI, and hemoglobin levels, was another objective factor analyzed in this study. We found that BMI was negatively correlated with CRF (standardized regression coefficient = 0.35), while PNI was positively correlated (standardized regression coefficient = 0.12). These findings are consistent with previous research demonstrating a high incidence of nutritional risk in preoperative HCC patients, which is strongly associated with CRF51. Therefore, peri-interventional nutritional management should be prioritized to mitigate CRF. A surprising finding was that married patients reported higher fatigue levels than unmarried patients, which contrasts with previous studies on transcatheter arterial embolization in HCC patients18. One possible explanation is that the majority of participants in this study were middle-aged men, who, as the primary breadwinners, may experience additional psychological and financial stress due to their illness. The inability to fulfill family responsibilities, coupled with the stigma and emotional burden of their disease, may exacerbate fatigue in this population. Thus, healthcare providers should consider the unique challenges faced by married HCC patients when developing CRF management strategies. Strengths and limitations The strengths of this study include its detailed description of CRF and associated factors in peri-interventional HCC patients, as well as the development of a pathway model to explore the effects of PA, positive mood, insomnia, and liver function on CRF. These findings provide valuable insights for clinicians to predict CRF risk and for researchers to design targeted interventions for advanced HCC patients. This paper has a few shortcomings. First, because we have only a single center and rely on self-reported data, recall bias may have a bearing on the study and it is not statistically possible to detect subtle associations. Second, we may have a selection bias. Patients with severe fatigue may not have been more likely to participate, and those without fatigue may not have been so motivated to enroll. Thirdly, we did not collect information on opioid use or dosage. Opioid analgesics are frequently prescribed for cancer pain and are known to induce or exacerbate fatigue. The absence of this variable may have led to an underestimation of certain contributors to CRF. Future studies should include detailed records of analgesic medications, especially opioids, to better control for their confounding effects. Finally, the study population was limited to HCC patients undergoing interventional therapy, and further research is needed to determine whether the proposed pathway model is applicable to other treatment modalities or cancer types. Future studies should employ longitudinal designs and objective measures to validate the effects of exercise and liver function on positive mood and CRF in advanced HCC. Conclusion In conclusion, this study provides the first comprehensive path analysis of factors associated with CRF in HCC patients during the peri-intervention period. The findings reveal a high prevalence and severity of CRF in this population. Path analysis identified hope as the strongest correlate of CRF, followed by self-efficacy, liver function, and insomnia. PA had no direct effect on CRF but operated indirectly through hope and self-efficacy, suggesting that psychological mechanisms may mediate the exercise-fatigue relationship in this population. These findings have several practical implications. Clinicians should prioritize routine CRF screening and targeted interventions that enhance hope and self-efficacy, optimize liver function, improve sleep quality, and correct nutritional deficits (as reflected by BMI and PNI). Addressing economic burden and marital status may also help mitigate fatigue risk. The mediating role of psychological factors suggests that physical activity programs may be more effective when combined with cognitive-behavioral strategies. The present study has several limitations that should be acknowledged. The cross-sectional design precludes causal inferences. The single-center setting and reliance on self-reported data may introduce selection and recall bias. The generalizability of the pathway model to other cancer types or treatment modalities remains to be determined. Moreover, opioid use—a potential confounder—was not assessed. Future longitudinal studies should incorporate objective measures of physical activity, sleep, and liver function, as well as detailed medication records, to validate these findings and establish causality. In summary, CRF is a highly prevalent and severe symptom in advanced HCC patients undergoing peri-intervention care. Multidimensional interventions that promote hope and self-efficacy, improve liver function and sleep, and address nutritional and financial needs may effectively reduce CRF burden. Healthcare providers should integrate these modifiable factors into routine clinical practice to enhance patient quality of life and clinical outcomes.