Constructing a neural network model based on tumor-infiltrating lymphocytes (TILs) to predict the survival of hepatocellular carcinoma patients

Abstract

Background: Hepatocellular carcinoma (HCC) is the most common primary liver cancer worldwide, and early pathological diagnosis is crucial for formulating treatment plans. Despite the widespread attention to pathology in the treatment of HCC patients, a large amount of information contained in pathological images is often overlooked.

Methods: We retrospectively collected clinical data and pathological slide images from (a) 331 HCC patients at Qingdao University Affiliated Hospital between January 2013 and December 2016 and (b) 180 HCC patients from The Cancer Genome Atlas (TCGA). After data screening, precise quantification of various cell types was achieved using QuPath software. Key factors related to the survival prognosis of pathologically confirmed HCC patients were identified through Cox regression and neural network models, and potential therapeutic targets were screened.

Results: Our study showed that tumour-infiltrating lymphocytes (TILs) had a protective effect. We quantified the TILs index by machine learning and built a neural network model to predict the prognostic risk of patients (ROC = 0.836 for training set ROC validation set). 95% CI [0.7688-0.896], and there was a significant difference in prognosis in the high-low risk group predicted by the model (p = 2.6e-18, HR = 0.18, 95% CI [0.12-0.27], and TNFSF4 was identified as a possible immunotherapy target.

Conclusion: This study included a total of 511 patients, divided into a training cohort of 331 cases (from Qingdao University Hospital between January 2013 and December 2016) and a validation cohort of 180 cases (TCGA). The results revealed that tumor-infiltrating lymphocytes (TILs) have a protective effect and successfully predicted the survival risk of liver cancer patients using machine learning and neural network technology. The discovery of TNFSF4 provides a new potential target for immunotherapy.

Keywords: Digital pathology; Hepatocellular carcinoma; Immune infiltration; Machine learning; Neural network model.

Figure 1. Workflow, software quantification interface, and Cox analysis.

(A) The flow chart of the study process. (B) Interface for quantifying TILs using QuPath. (C, D) Univariate and multivariate Cox analyses to evaluate independent prognostic factors of HCC. Tumor capsule, whether the tumor capsule is intact; liver capsule, whether the tumor invades the liver capsule; rupture, whether the tumor has ruptured; anatomical section, whether anatomical section was performed during surgery; MVI, microvascular invasion; PVI, portal vein tumor thrombus; LymphInvolvement, presence of lymphocytic infiltration around tumor nests; NR resection, whether it is radical resection; SALON, surgical method (laparoscopic or not); AFP, alpha-fetoprotein; PLT, platelets; INR, international normalized ratio; PT, prothrombin time; ALB, albumin; Tbil, total bilirubin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; Cr, creatinine; HPVO, hepatic portal vein occlusion; TBA, total bile acids.

Figure 2. TILs are independent prognostic factors in HCC patients.

(A) K-M curve of eTILs% stratified by median ALB. (B) K-M curve of eTILs% in the high ALB group. (C) K-M curve of eTILs% in the low ALB group. (D) K-M curve of eTILs% in patients with satellite nodules. (E) K-M curve of eTILs% in patients without satellite nodules. (F–I) K-M curves of the four features of TLSs.

Figure 3. Establishing a neural network model.

(A) Schematic diagram of the neural network model. (B) Performance comparison at different time cutoffs. (C) ROC curves of the validation group. (D) K-M curves of the high and low-risk groups.

Figure 4. The neural network model predicts patients in the high and low-risk groups.

(A) Volcano plot of differentially expressed genes between the two groups. (B) GO enrichment analysis of differentially expressed genes between the two groups. (C) Immune-related scores in the high and low-risk groups. (D) Differences in immune cells between the high and low-risk groups. (E) Selection of immune checkpoints in the high and low-risk groups.

Figure 5. Poor prognosis is linked to TNFSF4 high expression in HCC.

(A) Immunohistochemical analysis of TNFSF4 protein expression in human HCC tissue. (B) TNFSF4 protein expression in 116 patients with HCC was correlated with OS. (C) Differences in TNFSF4 mRNA expression between HCC tissues and normal tissues in patients with HCC in the TCGA database. (D) TNFSF4 mRNA expression differences between hepatocellular carcinoma tissue and adjacent normal tissue in paired samples from the TCGA database.