Butyrate, one of the major products of the gut microbiota, has been reported to play a significant antitumor role in various tumors, such as colon cancer, breast cancer, and gastric cancer, etc. [63,64,65,66]. Emerging studies suggested that butyrate inhibits the progression of HCC through multiple mechanisms [16,17]. Moreover, butyrate may impact the effectiveness of immunotherapy in solid tumors [7]. These studies demonstrated that butyrate might have great potential as a novel alternative for treating HCC. Nevertheless, the relevance of the butyric acid metabolic pathway to the prognosis and systemic therapies of HCC has not been comprehensively studied. Thus, we aimed to elucidate the role of the butyric acid metabolic pathway in the early diagnosis, prognosis assessment, and systemic treatment of HCC.

In our study, according to the expression levels of candidate BM-related genes, the HCC samples in the TCGA-LIHC cohort were grouped into A and B clusters, with most of the genes highly expressed in the latter. Patients in the B cluster showed worse prognoses and may be more likely to respond to immunotherapy. Our results suggested that BM-related subtype could be applied to predict prognoses and immunotherapy responses in patients with HCC.

To further reveal the underlying prognostic and molecular mechanisms of butyrate metabolism pathways for early diagnosis, prognoses, and systemic treatment, the BMGs were constructed by the LASSO Cox algorithm. The BMGs were comprised of eight genes, including LCAT, G6PD, SPP1, GLP1R, GAD1, MMP1, CCNA2, and MAPT. Recent evidence indicated that G6PD deficiency might inhibit the carcinogenesis, proliferation, and metastasis of HCC cells by upregulating cytochrome P450 oxidoreductase (POR), suggesting that G6PD may be applied as a biomarker for the treatment of HCC in the future [67]. CCNA2 facilitates cell cycle progression in HCC cells and is considered a molecular stimulator of HCC progression [68]. In addition, MMP1-regulated signaling pathways have been indicated to be involved in the metastasis and progression of HCC [69,70]. GAD1 has been reported to stimulate tumor cell invasion and metastasis by regulating β-catenin translocation and activating MMP7 [71]. GLP1R may stimulate the progression of HCC via the cAMP-PKA-EGFR-STAT3 axis [72]. It should be noted that the down-regulation of MAPT expression is a reliable predictive marker of drug sensitivity in tumor cells [73]. In our study, the network of these genes and particular transcription factors was constructed, which may provide innovative ideas for finding new targets for the treatment of HCC. Altogether, our study might establish a groundwork for upcoming studies, but the mechanisms of these genes in HCC still deserve further validation.

BMGs were established and validated in multiple datasets, including the TCGA-LIHC, ICGC-JP, GSE14520, GSE6764, and GSE9843 cohorts. We discovered that BMGs played a remarkable role in assessing the prognoses of HCC. Compared to the 33 published gene signatures in the PubMed database, BMGs may be more advantageous in evaluating the prognoses of HCC patients. Significantly, BMGs could distinguish HCC from other liver diseases during the early course of HCC based on GSE6764, GSE89377, and GSE49451 cohorts. More importantly, BMGs showed excellent predictive performances for evaluating the responses to diverse therapies, including immunotherapy, TACE therapy, ablation therapy, and drug sensitivity based on TCGA-LIHC, TCIA, TIDE, GSE91061, GSE115821, GSE104580, GSE143004, and GSE109211 cohorts. Our study demonstrated that patients in the high-risk group could gain benefits from immunologic and ablation therapy, while these patients in the low-risk group might be sensitive to TACE treatment and susceptible to “sorafenib”. These findings suggested that BMGs played a significant role in assessing the prognoses and multiple therapies of HCC. Furthermore, the most commonly mutated genes and potential molecular functions were distinct between these subgroups.

In the GSE6764, GSE89377, and GSE49541 cohorts, we discovered that BMGs could have the potential to serve the innovative biomarkers to identify a high-risk group of developing HCC, which was beneficial for the early diagnosis of HCC.

BMGs performed excellently in assessing the prognoses of HCC, which might be the independent risk factors for HCC prognoses. Importantly, compared to the other 33 published biomarkers, BMGs have demonstrated a superior potential to predict the prognosis of HCC. The nomogram based on the risk score was constructed, and all calibration, ROC, and c-index curves indicated the great clinical values of the nomogram in assessing the prognoses of HCC.

Based on BMGs, we might speculate on potential factors for varying prognoses between high- and low-risk groups, such as genetic mutation profile, mRNAsi, and immune features.

We hypothesized that genetic mutation profile might contribute to this outcome. Our study showed that a higher total mutation frequency was observed in the high-risk group. Notably, enhanced genomic instability is strongly associated with the prognosis of malignant neoplasms [74]. Somatic mutation analysis demonstrated that the most common genetic mutations in the high- and low-risk groups were TP53 and CTNNB1, respectively. Recent evidence suggests that HCC patients with a higher frequency of CTNNB1 mutations tend to present smaller and better-differentiated tumors [75]. Conversely, the phenomenon of vascular invasion in malignant tumors is closely associated with mutations in TP53 [75]. It is evident that genomic mutation profiles can result in distinct prognoses of high- and low-risk populations.Profiles of molecular functions may also lead to varied prognoses in patients in two subgroups as well. As shown in Figure 10A, the more abundant pathways involved in tumorigenesis and proliferation were concentrated in patients in the high-risk group. In addition, the GSEA analysis revealed that many invasion and proliferation pathways in cancers were enriched in the high-risk group, while several metabolic pathways clustered in the other group.Our results indicated significant differences in TME. In regard to immune cells, a greater amount of Th2 and Treg cells and a lower amount of NK cells were detected in the high-risk group. Immune functions also varied between subgroups. Our results suggested that there was a marked state of immunosuppression in the high-risk group, which contributed to tumor invasion and metastasis [76]. Altogether, immune features may also be responsible for the markedly varied prognoses of high- and low-risk groups.Furthermore, we identified enhanced mRNAsi in the high-risk group of patients. Moreover, the mRNAsi was positively associated with risk scores. It has been reported that higher mRNAsi represented tumor cells with more aggression [77]. Our study suggested that the stemness index profiles were another reason for the varied prognoses of the two subgroups.

According to BMGs in our study, we may implement more tailored treatment modalities for patients with HCC, especially regarding immunotherapy, TACE therapy, ablation therapy, and chemotherapy.

Immunotherapy is an effective therapy for some patients with HCC. However, one of the prerequisites to ensure the efficacy of immunotherapy is the ample expression of immune checkpoints in tumor tissues. In these years, immune checkpoint inhibitors (ICISs) targeting CTLA-4 and PDCD1 have achieved good efficacy for HCC patients [78]. Our results indicated that elevated expression levels of common checkpoints were enriched in the high-risk group. Additionally, TIDE and IPS scores confirmed that patients in the high-risk group could have a greater potential to obtain favorable outcomes from immunological therapy. Our study demonstrated that enhanced TIS, NFAG, and IFNAP in the high-risk group verified that patients in the high-risk group tended to respond to immunotherapy. Moreover, CTNNB1 mutations were more common in the low-risk group, and emerging evidence suggested that patients with higher CTNNB1 were more likely to be resistant to immunotherapy [79]. Our study initially showed that these regulatory genes could predict the efficacy of immunotherapy. Previous studies have indicated that butyrate supplementation could improve the effectiveness of immunotherapy in patients with malignant tumors. In the future, regulating BM related-genes to enhance the efficacy of immunotherapy in HCC patients might be a promising therapeutic direction.

TACE and ablation treatment are considered promising alternatives for patients with advanced HCC. However, in clinical practice, we found that not all patients with advanced HCC responded to TACE and ablation therapy. Our research initially addressed this challenge. Our study suggested that patients in the low-risk group might be more sensitive to TACE, while patients in the other group might respond to ablation therapy. Taken together, BMGs might be the effective biomarkers to assess the effectiveness of TACE and ablation therapy for HCC.

Chemotherapy remains the conventional treatment modality for a large proportion of patients with advanced HCC. However, faced with more than 300 chemotherapeutic agents, selecting the appropriate drug for special HCC patients has always been a puzzling problem for clinicians. Based on BMGs, sensitive chemotherapeutic drugs were screened for two subgroups, respectively. Our results revealed that patients in the high-risk group might have greater chances of gaining benefits from ‘paclitaxel’. Notably, we observed that tumors in the low-risk group of patients might be more vulnerable to “sorafila”. Altogether, based on BMGs, we could develop individualized drug treatment strategies, which may be critical in improving the prognoses of HCC.

Our study focused on the role of genes related to butyrate metabolism in the prognosis and systemic therapies of HCC in various datasets, which was innovative and few were previously reported. It was noteworthy that BMGs performed more favorably in evaluating the OS of HCC patients compared to 33 potential gene signatures in the PubMed database. In addition, our study is the first to systematically elucidate the role of gene signatures in the early detection of HCC, predicting responses to immunotherapy, drug sensitivity, TACE therapy, and ablative therapy in HCC, which was of great clinical promotion. Moreover, 16 datasets from different databases might increase the credibility of the findings. However, since these cohorts in this study were collected on various public platforms, the presence of tumor heterogeneity in patients included in different data sets is inevitable. We demonstrated that BMGs were superior to other gene signatures in predicting HCC prognoses in the TCGA-LIHC data set, without validation in other data sets, which is also a limitation of this study. The role of genes modulating butyrate metabolism in evaluating the prognoses and systemic therapies of HCC remains to be confirmed by prospective cohort studies and basic research in large samples.