An increasing body of evidence has suggested the close correlation of immune microenvironment with tumorigenesis and cancer progression [34–36]. By analyzing the immune landscape of HCC microenvironment, some researchers pointed out that the immune contexture could be a major prognostic indicator, and should not be disregarded to enhance the potential of precision treatments [37]. At present, immunotherapy has been widely recognized to treat a variety of cancers including HCC [38–40]. However, not all patients can benefit from it. Therefore, it is necessary to establish an IRG signature for survival prediction of HCC patients and enriching the effective population of cancer immunotherapy.
During the past years, genomics and bioinformatics have enabled the identification of molecular signatures. For example, several signatures have been identified for prognostic prediction based on lncRNA, miRNA, and mRNA [41, 42]. In this study, IRGPI was constructed by integrating the clinical information and transcriptomic data of HCC samples in TCGA cohort and GSE14520 cohort. A total of 329 DEIRGs were identified, of which the most relevant biological process and signaling pathway was “immune response” and “cytokine-cytokine receptor interaction”, respectively. This result was closely associated with immune, which was consistent with some existing literature reports [43]. Subsequently, Cox regression analysis and LASSO regression model identified 11 out of 81 prognosis-related IRGs, which were used to construct IRGPI, including NDRG1, FABP6, MAPT, HSP90AA1, CD320, CACYBP, BRD8, OSGIN1, NRAS, ISG20L2, and PSMD14. Among them, NDRG1 has been reported to be an essential molecule in controlling the metastasis and recurrence of HCC [44]. In addition, the deletion of CACYBP has also been reported to increase apoptosis of HCC cells [45], while the variants of OSGIN1 could reduce apoptosis and are associated with shorter survival [46]. Besides, knockdown and overexpression assays have demonstrated that PSMD14 could promote migration and invasion of HCC cells in vitro, and facilitate tumor growth and metastasis in vivo [47]. Although the direct association between the other seven genes and HCC has not been discovered, we think that the underlying correlations deserve further experimental validation.
In consideration of the importance of immune cell infiltration in tumors, CIBERSORT was further adopted for evaluating the relative proportion of 22 types of immune cells in every HCC specimen. Some evidence has indicated that the interplay between tumor and microenvironment plays a critical role in HCC progression and the probability of response to immunotherapies. Our study suggested that IRGPI was significantly and positively associated with the relative proportion of activated memory CD4 T cells and M0 macrophages, which are the only two types of immune cells significantly associated with OS. Some studies have shown that the selective loss or apoptosis of intrahepatic CD4+ T lymphocytes would promote hepatocarcinogenesis [48, 49].
The advent of immunotherapy has shed novel light on HCC treatment, of which ICIs have become a potentially effective treatment [6]. Targeting immune checkpoint molecules such as PD-1 and CTLA-4 could reinvigorate anti-tumor immunity [50]. Recently, nivolumab and pembrolizumab, two therapeutics against PD-L1/PD1, have been recently approved for subsequent-line therapy [51]. In order to predict the reactivity of ICIs, the relationship between IRGPI and IPS was explored in HCC patients. The analysis indicated that the low-risk group had higher IPS and IPS-CTLA4 scores, revealing that IRGPI has the potential to determine the specific HCC patients who are immunogenic and more responsive to ICIs. The predictive value of IRGPI on the response to ICIs provides a theoretical basis for the therapeutic selection of ICIs in clinical practice. Hopefully, this predictive model could assist to accelerate the pace of individualized cancer immunotherapy.
To further enhance the accuracy of prognostic prediction, we constructed and validated a nomogram by integrating IRGPI, age, gender, tumor status, tumor grade, pathological stage and T stage. Similarly, Ying et al. [52] combined inflammatory biomarkers with risk factors to form a nomogram, which could improve the accuracy for predicting clinical outcomes in CRC patients undergoing surgical resection. More importantly, these new prognostic tools could not only improve the accuracy of prognostic prediction, but also help to predict the specific survival risk of individual patients, which is of great significance in clinical practice [53].
There are several strengths in this study. Firstly, this signature was sufficiently validated and evaluated in multiple datasets, indicating the robustness and reliability of the signature. Secondly, comprehensive and in-depth researches were carried out in various aspects, including discussions on the correlation of IRGPI with the immune cells, IPS and TMB. Thirdly, a nomogram was further established for the quantitative calculation, which is conducive to clinical promotion and application. Nevertheless, several limitations still exist in our study. Thus, more HCC patients and validations are warranted to further test this signature by prospective studies in the future.