Machine Learning to Predict the Response to Lenvatinib Combined with Transarterial Chemoembolization for Unresectable Hepatocellular Carcinoma

Abstract

Background: Lenvatinib and transarterial chemoembolization (TACE) are first-line treatments for unresectable hepatocellular carcinoma (HCC), but the objective response rate (ORR) is not satisfactory. We aimed to predict the response to lenvatinib combined with TACE before treatment for unresectable HCC using machine learning (ML) algorithms based on clinical data.

Methods: Patients with unresectable HCC receiving the combination therapy of lenvatinib combined with TACE from two medical centers were retrospectively collected from January 2020 to December 2021. The response to the combination therapy was evaluated over the following 4-12 weeks. Five types of ML algorithms were applied to develop the predictive models, including classification and regression tree (CART), adaptive boosting (AdaBoost), extreme gradient boosting (XGBoost), random forest (RF), and support vector machine (SVM). The performance of the models was assessed by the receiver operating characteristic (ROC) curve and area under the receiver operating characteristic curve (AUC). The Shapley Additive exPlanation (SHAP) method was applied to explain the model.

Results: A total of 125 unresectable HCC patients were included in the analysis after the inclusion and exclusion criteria, among which 42 (33.6%) patients showed progression disease (PD), 49 (39.2%) showed stable disease (SD), and 34 (27.2%) achieved partial response (PR). The nonresponse group (PD + SD) included 91 patients, while the response group (PR) included 34 patients. The top 40 most important features from all 64 clinical features were selected using the recursive feature elimination (RFE) algorithm to develop the predictive models. The predictive power was satisfactory, with AUCs of 0.74 to 0.91. The SVM model and RF model showed the highest accuracy (86.5%), and the RF model showed the largest AUC (0.91, 95% confidence interval (CI): 0.61-0.95). The SHAP summary plot and decision plot illustrated the impact of the top 40 features on the efficacy of the combination therapy, and the SHAP force plot successfully predicted the efficacy at the individualized level.

Conclusions: A new predictive model based on clinical data was developed using ML algorithms, which showed favorable performance in predicting the response to lenvatinib combined with TACE for unresectable HCC. Combining ML with SHAP could provide an explicit explanation of the efficacy prediction.

Keywords: Shapley Additive exPlanation; hepatocellular carcinoma; lenvatinib; machine learning; transarterial chemoembolization; treatment response.

Figure 1

Flow chart of the study.

Figure 2

ROC curves for differentiating nonresponse and response using the machine learning algorithms. ROC, receiver operating characteristic; CART, classification and regression tree; AdaBoost, adaptive boosting; XGBoost, extreme gradient boosting; SVM, support vector machine; RF, random forest.

Figure 3

Importance matrix plot of the RF, AdaBoost, and XGBoost models. (A) Relative importance of the variables for segregation of the nonresponse group and response group calculated in the RF, AdaBoost, and XGBoost models. The variable importance is represented as a percentage of the highest value. (B) Box and jitter plots representing the distribution of the top 10 important features for distinguishing nonresponse and response. LDL, low-density lipoprotein; D-D, d-dimer; RBC, red blood cell; ALT, alanine aminotransferase; ALB, albumin; Mono, monocyte; TG, triglyceride; LMR, lymphocyte monocyte ratio; BMI, body mass index; TC, total cholesterol; APTT, activated partial thromboplastin time; INR, international normalized ratio; PT, prothrombin time; I-BIL, indirect bilirubin; ALBI, albumin–bilirubin; T-BIL, total bilirubin; AST, aspartate aminotransferase; CEA, car-cinoembryonic antigen; Hb, hemoglobin; A/G, albumin/globulin; GLOB, globulin; FIB, fibrinogen; PLT, platelet; γ-GTP, γ-glutamyl transpeptidase; HDL, high-density lipoprotein; AFP, alpha fetoprotein; TP, total protein; NLR, neutrophil lymphocyte ratio; Lymph, lymphocyte; BUN, blood urea nitrogen; PLR, platelet lymphocyte ratio; CA19-9, carbohydrate antigen 19-9; Scr, serum creatinine; Neut, neutrophil; ALP, alkaline phosphatase; WBC, white blood cell; NR, nonresponse; R, response.

Figure 4

SHAP plot of the RF model: (A) SHAP summary plot of the top 40 features of the RF model; (B) SHAP decision plot of the top 40 features of the RF model; (C) SHAP force plot for the explanation of the model prediction results with a response sample from the testing set; (D) SHAP force plot for the explanation of the model prediction results with a nonresponse sample from the testing set. LDL, low-density lipoprotein; ALT, alanine aminotransferase; RBC, red blood cell; D-D, d-dimer; ALB, albumin; TC, total cholesterol; TG, triglyceride; Mono, monocyte; LMR, lymphocyte monocyte ratio; BMI, body mass index; I-BIL, indirect bilirubin; AST, aspartate aminotransferase; APTT, activated partial thromboplastin time; INR, international normalized ratio; PT, prothrombin time; ALBI, albumin–bilirubin; T-BIL, total bilirubin; CEA, carcinoembryonic antigen; Hb, hemoglobin; A/G, albumin/globulin; GLOB, globulin; FIB, fibrinogen; HDL, high-density lipoprotein; γ-GTP, γ-glutamyl transpeptidase; PLT, platelet; TP, total protein; NLR, neutrophil lymphocyte ratio; AFP, alpha fetoprotein; Lymph, lymphocyte; BUN, blood urea nitrogen; PLR, platelet lymphocyte ratio; Neut, neutrophil; CA19-9, carbohydrate antigen 19-9; ALP, alkaline phosphatase; Scr, serum creatinine; WBC, white blood cell; SHAP, Shapley Additive exPlanation.

Figure 4

SHAP plot of the RF model: (A) SHAP summary plot of the top 40 features of the RF model; (B) SHAP decision plot of the top 40 features of the RF model; (C) SHAP force plot for the explanation of the model prediction results with a response sample from the testing set; (D) SHAP force plot for the explanation of the model prediction results with a nonresponse sample from the testing set. LDL, low-density lipoprotein; ALT, alanine aminotransferase; RBC, red blood cell; D-D, d-dimer; ALB, albumin; TC, total cholesterol; TG, triglyceride; Mono, monocyte; LMR, lymphocyte monocyte ratio; BMI, body mass index; I-BIL, indirect bilirubin; AST, aspartate aminotransferase; APTT, activated partial thromboplastin time; INR, international normalized ratio; PT, prothrombin time; ALBI, albumin–bilirubin; T-BIL, total bilirubin; CEA, carcinoembryonic antigen; Hb, hemoglobin; A/G, albumin/globulin; GLOB, globulin; FIB, fibrinogen; HDL, high-density lipoprotein; γ-GTP, γ-glutamyl transpeptidase; PLT, platelet; TP, total protein; NLR, neutrophil lymphocyte ratio; AFP, alpha fetoprotein; Lymph, lymphocyte; BUN, blood urea nitrogen; PLR, platelet lymphocyte ratio; Neut, neutrophil; CA19-9, carbohydrate antigen 19-9; ALP, alkaline phosphatase; Scr, serum creatinine; WBC, white blood cell; SHAP, Shapley Additive exPlanation.