Machine learning integrations develop an antigen-presenting-cells and T-Cells-Infiltration derived LncRNA signature for improving clinical outcomes in hepatocellular carcinoma

Abstract

As a highly heterogeneous cancer, the prognostic stratification and personalized management of hepatocellular carcinoma (HCC) are still challenging. Recently, Antigen-presenting-cells (APCs) and T-cells-infiltration (TCI) have been reported to be implicated in modifying immunology in HCC. Nevertheless, the clinical value of APCs and TCI-related long non-coding RNAs (LncRNAs) in the clinical outcomes and precision treatment of HCC is still obscure. In this study, a total of 805 HCC patients were enrolled from three public datasets and an external clinical cohort. 5 machine learning (ML) algorithms were transformed into 15 kinds of ML integrations, which was used to construct the preliminary APC-TCI related LncRNA signature (ATLS). According to the criterion with the largest average C-index in the validation sets, the optimal ML integration was selected to construct the optimal ATLS. By incorporating several vital clinical characteristics and molecular features for comparison, ATLS was demonstrated to have a relatively more significantly superior predictive capacity. Additionally, it was found that the patients with high ATLS score had dismal prognosis, relatively high frequency of tumor mutation, remarkable immune activation, high expression levels of T cell proliferation regulators and anti-PD-L1 response as well as extraordinary sensitivity to Oxaliplatin/Fluorouracil/Lenvatinib. In conclusion, ATLS may serve as a robust and powerful biomarker for improving the clinical outcomes and precision treatment of HCC.

Survival Prediction Landscape: An In-Depth Systematic Literature Review on Activities, Methods, Tools, Diseases, and Databases

“…A closer look at the clinical features across diverse diseases reveals a consistent set of fundamental demographic features i.e., age and gender which are prevalent in nearly all studies 85,86,91,111,112,115 . Beyond demographic features, diseasespecific features also play critical role for disease-specific survival prediction.…”

“…Table 8 provides information about 44 diseases and the corresponding survival prediction algorithms utilized in these diseases. A deeper analysis of Table 8 shows that Cox-PH and lasso Cox-PH models have been extensively utilized for disease specific survival prediction i.e., ASCVD 29,111 , bladder cancer 40,82 , colorectal cancer [74][75][76][77] , hepatocellular carcinoma 43,86,87 , ovarian cancer [88][89][90]103 , lung adenocarcinoma 101 , heart failure 118 , HER2-negative metastatic breast cancer 67 , pancreatic cancer 26,71 , trauma 120 , nasopharyngeal carcinoma 66 , triple-negative breast cancer 68 , lymphoma 85 , breast cancer 40,81,82 , ovarian cancer [88][89][90]103 , and lower-grade glioma 80 , cardiovascular disease 112,[114][115][116][117] , invasive ductal carcinoma 70 , liver transplantation 119 , gastric cancer 42 , lung cancer 27 , esophageal squamous cell carcinoma 79 , glioma 69 , and liver cancer …”

“…Treatment-related features such as chemotherapy, radiotherapy, and immunotherapy, are particularly evident in cancer subtypes specific studies which reflect the profound influence of therapeutic interventions on survival outcomes 86,98 . Moreover, the recurrent inclusion of lifestyle and comorbidity factors ranging from smoking history and BMI to hypertension and diabetes across multiple diseases underlines their pervasive impact on prognostic modeling 101,111 .…”

“…These methods are broadly categorized into five categories, namely supervised methods, incorporating L1 regularized Cox regression 29 , RSF algorithm 29 , Cox regression 103 , least absolute shrinkage and selection operator (lasso) regression 120 , cascaded Wx 105 , recursive feature elimination 38 , Boruta 31 , Akaike information criterion (AIC) regression 114 , variance 72 , lasso analysis 40 , multivariate regression 40 , Chi-squared 118 , mutual information 118 , and ANOVA 39,118 . Additionally, Network based methods include network based stratification (NBS) 83 , weighted correlation network analysis (WGCNA) 86 , canonical correlation analyses (CCA) 67 , patient similarity networks 38 , and neighborhood component analysis (NCA) 23 . Dimensionality reduction methods include non-negative matrix factorization (NMF) 40 , autoencoders (AEs) 28 , variational autoencoders (VAEs) 43 , principal component analysis (PCA) 39 , and dominant effect of the cancer driver genes (DEOD) 75,132 .…”

“…A comprehensive analysis of feature engineering methods across a range of disease-specific survival prediction studies unveils that supervised methods, such as Cox regression, L1 regularized Cox regression, and RSF algorithm, have been prevalent in diseases like ASCVD, trauma, and ovarian cancer 103,120 . On the other hand, network based methods including NBS and WGCNA, have been applied in diseases like KIRP, and hepatocellular carcinoma, which shows the significance of network structures in certain medical contexts 86 . Univariate analyses, including ANOVA, chi-squared, and univariate Cox regression, have been prevalent in diseases such as pancreatic cancer and heart failure, underscoring the significance of statistical testing in identifying relevant features 71,118 .…”