General information of the research participants

This study collected data from 210 patients who met the inclusion criteria between January 2019 and December 2023. Exclusions included five patients transferred from other hospitals who immediately underwent TACE procedures and who had more than 20% preoperative-related data missing in electronic medical records; three patients experiencing HCC rupture with haemorrhage who received massive blood product transfusions upon admission, rendering laboratory indicators unable to reflect actual nutritional status; and eight patients lost to follow-up following discharge without documented malnutrition during postoperative hospitalisation. Ultimately, data from 194 patients were included for statistical analysis (Fig. 1).

Fig. 1

Recruitment flowchart of research subjects

Of the total 194 patients with HCC who underwent TACE, 148 were men (76.29%) and 46 were women (23.71%). The patients’ age range was 27–85 years, with an average age of 64.64 ± 9.892 years. Regarding marital status, 190 patients were married (97.94%) and four were unmarried (2.06%). The distribution of educational levels was as follows: 85 patients (43.81%) had an education level of primary school or below, 84 (43.3%) had a middle school education and 25 (12.89%) had a college or higher education.

The medical insurance situation indicated that there were 116 patients (59.79%) with medical insurance and 78 (40.21%) without medical insurance. The distribution of TNM stages was as follows: stage I, 31 cases (15.98%), stage II, 46 cases (23.71%), stage III, 54 cases (27.84%) and stage IV, 63 cases (32.47%). In terms of CP classification, 77 patients (39.69%) were at grade A, 85 (43.81%) were at grade B and 32 (16.49%) were at grade C.

In terms of nutrition-related indicators, the BMI range of the study participants was 18.59–35.16 kg/m2, with an average of 22.85 ± 2.981 kg/m2; the L3-PMI range was 1.625–10.229 cm2/m2, with an average of 5.36 ± 1.109 cm2/m2; ALB levels ranged from 20 to 48.8 g/L, with an average of 35.71 ± 4.047 g/L; and Hb levels ranged from 64 to 171 g/L, with an average of 124.76 ± 21.261 g/L.

Based on the diagnostic criteria for malnutrition, patients were divided into a malnourished group and a non-malnourished group. There was no statistically significant difference in the general data between the two groups of patients (P > 0.05). Among the 194 patients, 70 experienced malnutrition, with an incidence rate of 36.08%.

Univariate analysis

Using the patient’s baseline risk factors as independent variables and the occurrence of malnutrition as the dependent variable, a univariate analysis was conducted. The results showed that age, BMI, CP grade, TNM stage, L3-PMI, ALB and Hb were all significantly associated with the occurrence of malnutrition following TACE (P < 0.05), indicating that these may be important factors affecting the patient’s nutritional status post-surgery.

In the malnutrition group, the average age of patients was 67.84 ± 10.670 years, which was significantly higher than the 62.12 ± 9.832 years in the non-malnutrition group (P < 0.001), suggesting that older individuals are more prone to malnutrition. The mean BMI was significantly lower in the malnutrition group (22.715 ± 3.232 kg/m2) than in the non-malnutrition group (23.64 ± 2.939 kg/m2) (P = 0.044), indicating that a lower BMI may be associated with the occurrence of postoperative malnutrition.

In terms of CP classification, the incidence of malnutrition was highest in patients classified as grade C (96.88%), whereas it was lowest in patients classified as grade A (5.19%), with a statistically significant difference (P < 0.001), indicating that patients with poorer liver function are more likely to experience malnutrition. Moreover, TNM staging is also significantly associated with the occurrence of malnutrition, with the incidence of malnutrition in patients at stage IV (55.56%) being much higher than in those at stage I (12.90%) (P < 0.001), suggesting that the higher the degree of tumour progression is, the poorer the nutritional status of the patients may be.

In terms of laboratory indicators, the mean L3-PMI (4.16 ± 0.973 cm2/m2) was significantly lower in the malnutrition group than in the non-malnutrition group (6.58 ± 1.219 cm2/m2) (P < 0.001), suggesting that lower muscle mass may increase the risk of malnutrition. The ALB level was significantly lower in the malnutrition group (33.76 ± 4.063 g/L) than in the non-malnutrition group (37.17 ± 4.347 g/L) (P < 0.001), indicating that hypoalbuminaemia may be closely related to postoperative malnutrition. Similarly, the Hb level in the malnutrition group (114.76 ± 21.852 g/L) was significantly lower than that in the non-malnutrition group (131.19 ± 21.281 g/L) (P < 0.001), suggesting that anaemia may be one of the risk factors for malnutrition.

The results of univariate analysis indicated that patients with older age, lower BMI, higher CP grade, later TNM stage, lower L3-PMI, lower ALB and lower Hb are more prone to malnutrition, suggesting that these factors may be important predictors of malnutrition in patients with HCC undergoing TACE. The specific results of the univariate analysis are presented in Table 1.

Table 1 Comparison of different clinical characteristics and univariate analysis results between malnourished and non-malnourished groupsConstruction of a nutritional risk prediction model for patients undergoing transarterial chemoembolisation for hepatocellular carcinoma

Variables with statistical significance (P < 0.05) in the univariate analysis were included in the multivariate logistic regression analysis to further screen for independent risk factors for malnutrition in patients undergoing TACE for liver cancer. The results of the regression analysis indicated that age, CP grade, L3-PMI, ALB and Hb were all independent influencing factors (Table 2).

Table 2 Results of Logistic Regression Analysis for Malnutrition Risk Factors

For every additional year of age, the risk of malnutrition in patients increases by 6.8% (odds ratio [OR] = 1.068, 95%CI: 1.012–1.128, P = 0.017). Patients with CP grade B and C have a significantly higher risk of malnutrition, especially those with grade C, whose OR value was as high as 11.780 (95%CI: 4.071–34.082, P < 0.001), indicating that the poorer the liver function is, the higher the risk of malnutrition following surgery.

Patients with lower L3-PMI levels are more prone to malnutrition (OR = 0.062, 95%CI: 0.017–0.230, P = 0.036), indicating that sarcopenia may be an important influencing factor for malnutrition following TACE surgery for liver cancer. Lower ALB levels (OR = 0.804, 95%CI: 0.726–0.890, P < 0.001) and lower Hb levels (OR = 0.971, 95%CI: 0.943–1.000, P = 0.048) are both closely related to malnutrition, suggesting that hypoalbuminaemia and anaemia may increase the risk of postoperative malnutrition.

A malnutrition risk prediction model for patients with HCC undergoing TACE treatment was constructed based on logistic regression analysis, with the regression equation as follows:

Logit(P) = 3.388–0.030 × Hb + 0.066 × Age + 2.466 × CP + 0.218 × ALB − 2.774(L3-PMI).

Evaluation of the nutritional risk prediction model for patients undergoing transarterial chemoembolisation for hepatocellular carcinoma

To evaluate the predictive ability of the constructed model, the HL goodness-of-fit test was first conducted. The results showed a likelihood ratio chi-square value of 8.584 (df = 8, P = 0.378), with a P-value of > 0.05, indicating that the model’s goodness-of-fit is satisfactory and the match between the predicted results and the actual observed outcomes is high (Fig. 2).

Fig. 2

ROC Curve Analysis of Regression Models and Variables in the Models

Further adoption of ROC curve analysis was used to evaluate the predictive performance of the model. The results showed that the area under the ROC curve (AUC) of the entire model was 0.958 (95%CI: 0.933–0.982, P < 0.001), indicating that the model has high predictive performance.

The ROC curve analysis was performed on the variables in the logistic regression model, and the results indicated that the AUC value for CP classification ranged between 0.85 and 0.95 (AUC = 0.853, 95%CI: 0.798–0.908, P < 0.001), indicating a relatively good individual predictive efficacy. The AUC values for age and TNM staging were between 0.5 and 0.7 (AUC = 0.669 and 0.694, P < 0.001), suggesting lower predictive efficacy. The AUC values for Hb, BMI, L3-PMI and ALB were all < 0.5 (P < 0.001), indicating that these individual indicators have lower predictive efficacy and are not as effective as the comprehensive model in predicting outcomes.

The results of the ROC curve analysis for the patients are shown in Table 3. The optimal cut-off value is 0.41, at which point the model’s sensitivity is 89.2% and the specificity is 92.4%, indicating that the model can accurately predict the risk of malnutrition in patients with liver cancer following TACE surgery. Additionally, this study calculated the Youden index as 0.814, further confirming that the model has good comprehensive discriminant ability in predicting malnutrition.

Table 3 Subject working curve (ROC curve) analysis results

In summary, the malnutrition risk prediction model for patients with HCC undergoing TACE surgery constructed in this study performs well in terms of model fit and predictive efficacy, and can provide a certain reference value for clinical practice.