HCC is the most common primary malignancy of the liver[3]. In recent years, scientists have made significant efforts to uncover the etiology and mechanism of HCC, and a large number of detection methods and treatment strategies have emerged, but the 5-year survival rate of HCC patients is still low[4]. Therefore, it is necessary to develop effective risk stratification and individualized treatment plan. Numerous studies have shown that abnormal PPARs are involved in the tumorigenesis and progression of various tumors, including skin cancer, colon cancer, breast cancer, and lung adenocarcinoma[15, 16, 18–20]. Despite the important roles of PPARs in HCC, studies on the association between PPARs and patient prognosis are still scarce. In the present study, we divided HCC patients into two molecular subtypes based on the expression levels of PPARGs, which had significant differences in prognosis and metabolic activity. Immune analysis indicated that HCC patients in cluster 1 had disordered immune status, higher immune score and higher proportion of immune cell infiltration than cluster 2. Further functional analyses demonstrated that abnormal expression of PPARs affects metabolic activity and tumor immunity. Moreover, we constructed a prognostic risk model based on four PPARGs (TTC33, TMEM135, TALDO1, and TXNIP), which could accurately predict the prognosis of HCC patients. Finally, we integrated clinical features and risk score to establish a nomogram, which further improved the accuracy of predicting the prognosis of HCC patients. Our findings may facilitate the development of targeted therapy for HCC and guide the formulation of rational clinical treatment decisions.
In this study, we firstly divided HCC patients into two molecular subgroups via consensus clustering analysis based on the PPARGs expression matrix of TCGA-LIHC, with distinct differences in tumor progression and clinical outcomes. Subsequently, we explored the function of PPARs in HCC by immunological and functional analyses of two molecular subgroups. Tumor progression is associated with changes in surrounding stroma, since TIME determines the prognosis of patients to a certain extent, with immune cells are the key components of TIME[21, 22]. The ESTIMATE algorithm can infer the proportion of immune and stromal cells in tumors according to gene expression values. The immune score obtained via the ESTIMATE algorithm shows the immune components of each tumor sample, which can reflect TIME more objectively[23]. A previous study by Xiang and Hu et al. identified that higher immune/stromal/estimate scores were significantly associated with better survival benefit in HCC patients[24, 25]. However, some studies had also pointed out that the immune score was higher in the subgroup with poor prognosis of HCC[26]. Therefore, we cannot judge the prognosis of HCC patients simply based on the immune score, it can only indicate that the immune status of HCC patients is disordered. In this study, we used the ESTIMATE algorithm to assess the difference in TIME between the identified molecular subgroup. Our results presented that HCC patients in cluster 1 had faster disease progression and higher immune score than cluster 2. In addition, we applied the other methods, MCP-counter and ssGSEA, to assess the immune status of two molecular subgroups. The results of MCP-counter analysis showed that the number of seven out of ten immune cells was significantly increased in cluster 1, which was consistent with the results of ESTIMATE analysis. These results indicated that the immune status of cluster 1 was up-regulated. The ssGSEA analysis summarized the proportion of 28 immune cells in HCC samples, and we found that the proportion of 16 immune infiltrating cells was significantly increased, indicating that cluster 1 was in a correspondingly high immune state, further proving the reliability of the ESTIMATE and TIMER results. Taken together, we can reasonably assume that the disturbance of the immune system is closely related to the prognosis of HCC patients.
Next, we explored the underlying biological mechanisms of the two molecular subgroups by functional analysis. GO, KEGG and PPI analyses synergistically suggested that aberrant metabolic regulation may mediate the function of PPARs in HCC formation and progression based on the identified DEGs. However, the exact relationship between PPARs and metabolic dysregulation remains unknown. Therefore, we used GSEA and GSVA analyses to further research the in-depth mechanisms. GSVA analysis calculates the activity of signaling pathways in each sample based on gene expression matrix and can estimate differences between groups. The results of GSVA analysis identified that various metabolic pathways were significantly inhibited in cluster 1. GSEA analysis is able to integrate gene expression information to directly elucidate the expression trends of signaling pathway gene sets in different groups. In the present study, the GSEA results also indicated that the corresponding metabolic level was low in cluster 1. These results explain that the abnormality of PPARs significantly alters the metabolic level of HCC patients, affecting the clinical outcomes of patients.
Based on the above research results, we can reasonably infer that the abnormality of PPARs will cause the disorder of metabolic level and TIME in patients, and ultimately lead to the poor prognosis of HCC. As previously mentioned, PPAR is important mediator closely related to tumor metabolism[27]. Aberrant PPAR signaling pathway in tumors has gained widespread attention over the past few decades. The PPAR pathway regulates energy metabolism and inflammatory response in the body, which exerts anti-fibrotic and anti-inflammatory effects in various diseases, including tumors, autoimmune diseases, hepatic steatosis, and type 2 diabetes (T2D)[28]. Activation of PPAR-α can induce apoptosis and tumor cell death, preventing uncontrolled tumor expansion and inflammatory responses. Meanwhile, PPAR-α acts as an inhibitor of colon carcinogenesis in mouse, which expression is downregulated in tumors[29]. PPAR-α deficient mice promote methylation of tumor suppressor genes P21 and p27 by improving the expression of DNMT1 and PRMT6[30]. Moreover, Rajarajan et al. demonstrated that PPAR-α knockdown down-regulated the proliferative potential of leptin-induced breast cancer cells[18]. In a breast cancer study, PPAR-γ activation induced terminal differentiation of cancer cells into adipocytes and promoted their apoptosis accompanied by up-regulation of C/EBP-β expression, inhibiting tumor cell proliferation[31]. In addition, PPAR-γ activation significantly reduces the invasive and metastatic abilities of human gastric adenocarcinoma by down-regulating the ERK-signaling pathway[32]. In terms of metabolism, activation of PPAR-γ can improve glucose and fat metabolism, demonstrating the high energy demand for prostate cancer proliferation, migration and invasion[33]. Wang et al. showed that ectopic expression of PPAR-δ/β in MCF-7 cells increased their migratory ability and resistance to endoplasmic reticulum stress conditions such as hypoxia and low glucose, inducing its colonization in mice lung[34]. In non-small cell lung cancer, the activation of PPAR-δ/β promoted the proliferative activity of cells, while the low expression of PPAR-δ/β decreased cell apoptosis[20]. These observations are consistent with the role for PPARs in promoting growth, apoptosis resistance, inflammation and angiogenesis in tumor. The above studies have shown that the abnormality of PPAR changes the metabolic activity of tumor cells to some extent, which in turn affects the prognosis of patients.
The role of PPARs in immune cells has been widely investigated during the last years. As key regulators of metabolism, PPARs direct the differentiation, proliferation and death of various immune cells. PPAR-γ has been proven to regulate the polarization, maturation, epigenetics and metabolism of macrophages[35]. Meanwhile, activation of PPAR-γ selectively inhibits Th17 cell differentiation in mouse CD4 + T cells[36]. Moreover, a novel PPAR-α antagonist (IS001) can increase IFN-γ secretion from NK cells, CD4 + T cells, and CD8 + T cells[37]. PPAR-β/δ can affect T cell development and function. PPAR-β/δ may enhance the proliferative ability of TCR-β selected thymocytes and the growth of peripheral CD4 + T cells by modulating the metabolic program of thymocytes[38]. These studies have revealed that abnormal PPAR signaling pathway leads to alter in immune cell status of TIME, which determines the clinical outcome of HCC patients.
To further validate the prognostic value of PPARGs in HCC, we constructed a prognostic risk model based on the expression profile of PPARGs, and verified via ICGC cohort. In this study, four genes were incorporated in our risk model (TTC33, TMEM135, TALDO1, and TXNIP), of which three genes had been shown to be closely associated with tumorigenesis and progression. The function of tetratricopeptide repeat domain 33 (TTC33) in tumors is still unclear, and there are few studies on TTC33. Matsumoto et al. found that the combined PRKAA1-TTC33 gene was expressed in the non-Hodgkin B-cell lymphoma (B-NHL) cell line KPUM-UH1, which may be a lymphoma-specific feature[39]. TMEM135 is a transmembrane protein that integrates biological processes involved in fat metabolism and energy consumption[40]. A multicenter study identified TMEM135 as a novel breast cancer gene by whole-genome massively parallel sequencing analysis of BRCA1 mutant oestrogen receptor-negative and -positive breast cancers[41]. Yu et al. found that the majority of TMEM135-CCDC67 transcript-positive prostate cancer patients experienced recurrence, distant metastasis and death after radical prostatectomy[42]. Transaldolase 1 (TALDO1) is a key enzyme in the pentose phosphate pathway. Previous studies have demonstrated that TALDO1 plays a critical role in accelerating cell proliferation by providing R5P for nucleic acid synthesis and NADPH for cell survival, particularly under stressful conditions. In addition, multiple studies have proven the expression of TALDO1 is associated with the prognosis of bladder cancer, prostate cancer, HCC, and ovarian cancer[43–46]. TXNIP has important functions in regulating mitochondrial function, inducing apoptosis, inhibiting growth and metastasis. In addition, it also plays a critical role in promoting natural killer cell development, arresting the cell cycle, regulating glucose metabolism and inflammatory signaling. Meanwhile, TXNIP is an essential regulator of glucose and lipid metabolism. Liang et al. identified that circDCUN1D4/HuR/TXNIP may form an RNA-protein ternary complex to suppress lung adenocarcinoma metastasis and glucose metabolism[47]. Numerous studies have shown that TXNIP is a potent tumor suppressor gene in breast, kidney, prostate and thyroid cancers, which is involved in metabolic reprogramming and oxidative stress of tumor cells[48]. Moreover, in our study, KM survival analysis and ROC analysis identified that the established risk model exhibited effective predictive prognostic performance for HCC patients, both in the training cohort and the validation cohort. Significantly lower stromal score, immune score and estimate score were associated with poorer prognosis. Furthermore, multivariate Cox survival analysis and subgroup analysis showed that the risk model could independently predict the prognosis of HCC regardless of the interference of confounding factors such as age, gender, tumor grade, TNM stage, T stage, N stage, and M stage. Ultimately, a nomogram that integrating risk score and clinical characteristics was constructed and calibrated, which presented excellent properties in predicting survival of HCC patients. All results suggest that PPARGs have a significant role in predicting prognosis of HCC, which are associated with abnormal metabolism and TIME.
In recent years, chemotherapy, molecular targeted therapy and immunotherapy of tumors have achieved significant advancement, but the prognosis of HCC patients is still poor. Therefore, it is extremely important to find an effective method that divides patients into different subgroups according to their prognostic risk score, and to formulate reasonable individualized and targeted treatment plans. Bioinformatics analysis based on RNA sequence is a feasible method for risk stratification and target identification. Although many scholars have constructed HCC risk models based on glucose metabolism, ferroptosis, tumor microenvironment, immune cell infiltration, and apoptosis, our study shows unique advantages compared with previous studies. First, our work focused on abnormal PPARs in HCC patients, and identified two molecular subgroups with different prognosis and immune status via consensus clustering approach. Subsequently, we explored the biological mechanisms of HCC based on the results of consensus clustering, and partially elucidated the underlying mechanisms. Meanwhile, we also demonstrated the effects of PPARs on metabolism, TIME and prognosis. Furthermore, our study incorporates the TCGA and ICGC databases, which contains more samples than previous studies. Our work will provide excellent theoretical guidance for in-depth research on the mechanism of HCC. Not only that, the results of this study can promote the research progress of HCC targeted therapy and help clinicians to formulate more rational treatment strategies.
Of course, there were still some flaws in our study, which should be addressed when generalizing the conclusions. First of all, our results were obtained via bioinformatics analysis, and had not been further verified by in vitro and in vivo experiments. Second, the data we used were available from public open databases and were not validated with our own research cohort. The low-level of evidence nature of retrospective studies still remained, and more prospective studies were needed to further confirm the prognostic value of PPARGs in HCC.