MetaPredictor: in silico prediction of drug metabolites based on deep language models with prompt engineering.

Metabolic processes can transform a drug into metabolites with different properties that may affect its efficacy and safety. Therefore, investigation of the metabolic fate of a drug candidate is of great significance for drug discovery. Computational methods have been developed to predict drug metabolites, but most of them suffer from two main obstacles: the lack of model generalization due to restrictions on metabolic transformation rules or specific enzyme families, and high rate of false-positive predictions. Here, we presented MetaPredictor, a rule-free, end-to-end and prompt-based method to predict possible human metabolites of small molecules including drugs as a sequence translation problem. We innovatively introduced prompt engineering into deep language models to enrich domain knowledge and guide decision-making. The results showed that using prompts that specify the sites of metabolism (SoMs) can steer the model to propose more accurate metabolite predictions, achieving a 30.4% increase in recall and a 16.8% reduction in false positives over the baseline model. The transfer learning strategy was also utilized to tackle the limited availability of metabolic data. For the adaptation to automatic or non-expert prediction, MetaPredictor was designed as a two-stage schema consisting of automatic identification of SoMs followed by metabolite prediction. Compared to four available drug metabolite prediction tools, our method showed comparable performance on the major enzyme families and better generalization that could additionally identify metabolites catalyzed by less common enzymes. The results indicated that MetaPredictor could provide a more comprehensive and accurate prediction of drug metabolism through the effective combination of transfer learning and prompt-based learning strategies.

DBPP-Predictor: a novel strategy for prediction of chemical drug-likeness based on property profiles.

Evaluation of chemical drug-likeness is essential for the discovery of high-quality drug candidates while avoiding unwarranted biological and clinical trial costs. A high-quality drug candidate should have promising drug-like properties, including pharmacological activity, suitable physicochemical and ADMET properties. Hence, in silico prediction of chemical drug-likeness has been proposed while being a challenging task. Although several prediction models have been developed to assess chemical drug-likeness, they have such drawbacks as sample dependence and poor interpretability. In this study, we developed a novel strategy, named DBPP-Predictor, to predict chemical drug-likeness based on property profile representation by integrating physicochemical and ADMET properties. The results demonstrated that DBPP-Predictor exhibited considerable generalization capability with AUC (area under the curve) values from 0.817 to 0.913 on external validation sets. In terms of application feasibility analysis, the results indicated that DBPP-Predictor not only demonstrated consistent and reasonable scoring performance on different data sets, but also was able to guide structural optimization. Moreover, it offered a new drug-likeness assessment perspective, without significant linear correlation with existing methods. We also developed a free standalone software for users to make drug-likeness prediction and property profile visualization for their compounds of interest. In summary, our DBPP-Predictor provided a valuable tool for the prediction of chemical drug-likeness, helping to identify appropriate drug candidates for further development.

XGraphCDS: An explainable deep learning model for predicting drug sensitivity from gene pathways and chemical structures.

Cancer is a highly complex disease characterized by genetic and phenotypic heterogeneity among individuals. In the era of precision medicine, understanding the genetic basis of these individual differences is crucial for developing new drugs and achieving personalized treatment. Despite the increasing abundance of cancer genomics data, predicting the relationship between cancer samples and drug sensitivity remains challenging. In this study, we developed an explainable graph neural network framework for predicting cancer drug sensitivity (XGraphCDS) based on comparative learning by integrating cancer gene expression information and drug chemical structure knowledge. Specifically, XGraphCDS consists of a unified heterogeneous network and multiple sub-networks, with molecular graphs representing drugs and gene enrichment scores representing cell lines. Experimental results showed that XGraphCDS consistently outperformed most state-of-the-art baselines (R2 = 0.863, AUC = 0.858). We also constructed a separate in vivo prediction model by using transfer learning strategies with in vitro experimental data and achieved good predictive power (AUC = 0.808). Simultaneously, our framework is interpretable, providing insights into resistance mechanisms alongside accurate predictions. The excellent performance of XGraphCDS highlights its immense potential in aiding the development of selective anti-tumor drugs and personalized dosing strategies in the field of precision medicine.

CAVIN2/SDPR Functioned as a Tumor Suppressor in Lung Adenocarcinoma from Systematic Analysis of Caveolae-Related Genes and Experimental Validation.

Background: Caveolae-Related Genes include caveolins and cavins, which are the main component of the fossa and, play important roles in a variety of physiological and pathological processes. Although increasing evidence indicated that caveolins (CAVs) and cavins (CAVINs) are involved in carcinogenesis and progression, their clinical significance and biological function in lung cancer are still limited. Methods: We investigated the expression of CAVs and CAVINs at transcriptional levels using Oncomine and Gene Expression Profiling Interactive Analysis. The protein and mRNA expression levels of CAVs and CAVINs were determined by the human protein atlas website and our surgically resected samples, respectively. The clinical value of prognostic prediction based on the expression of CAVs and CAVINs was also assessed. cBioPortal, GeneMANIA and STRING were used to analyze the molecular characteristics of CAVs and CAVINs in lung adenocarcinoma (LUAD) comprehensively. Finally, we investigated the effect of CAVIN2/SDPR (serum deprivation protein response) on LUAD cells with biological experiments in vitro. Results: The expression of CAV1/2 and CAVIN1/2/3 were significantly downregulated in LUAD and lung squamous cell carcinoma (LUSC). The patients with high expression of CAV1, CAV2, CAV3, CAVIN1 and CAVIN2/SDPR were tightly correlated with a better prognosis in LUAD, while no statistical significances in LUSC. Further, our results found that CAVIN2/SDPR can be identified as a prognostic biomarker independent of other CAVINs in patients with LUAD. Mechanically, the overexpression of CAVIN2/SDPR inhibited cell proliferation and migration owing to the cell apoptosis induction and cell cycle arrest at S phase in LUAD cells. Conclusions: CAVIN2/SDPR functioned as a tumor suppressor, and was able to serve as prognostic biomarkers in precision medicine of LUAD. Mechanically, overexpression of CAVIN2/SDPR inhibited cell proliferation by inducing cell apoptosis and S phase arrest in LUAD cells.

Thioredoxin facilitates hepatocellular carcinoma stemness and metastasis by increasing BACH1 stability to activate the AKT/mTOR pathway.

Thioredoxin (TXN) is essential for preserving balance and controlling the intracellular redox state. Most studies have focused on the function of TXN in redox reactions, which is critical for tumor progression. Here, we showed that TXN promotes hepatocellular carcinoma (HCC) stemness properties in a non-redox-dependent manner, which has rarely been reported in previous studies. TXN exhibited upregulated expression in human HCC specimens, which was associated with a poor prognosis. Functional studies showed that TXN promoted HCC stemness properties and facilitated HCC metastasis both in vitro and in vivo. Mechanistically, TXN promoted the stemness of HCC cells by interacting with BTB and CNC homology 1 (BACH1) and stabilized BACH1 expression by inhibiting its ubiquitination. BACH1 was positively correlated with TXN expression and was significantly upregulated in HCC. In addition, BACH1 promotes HCC stemness by activating the AKT/mammalian target of rapamycin (mTOR) pathway. Furthermore, we found that the specific inhibition of TXN in combination with lenvatinib in mice significantly improved the treatment of metastatic HCC. In summary, our data demonstrate that TXN plays a crucial role in HCC stemness and BACH1 plays an integral part in regulating this process by activating the AKT/mTOR pathway. Thus, TXN is a promising target for metastatic HCC therapy.

In Silico Prediction of Metabolic Reaction Catalyzed by Human Aldehyde Oxidase.

Aldehyde oxidase (AOX) plays an important role in drug metabolism. Human AOX (hAOX) is widely distributed in the body, and there are some differences between species. Currently, animal models cannot accurately predict the metabolism of hAOX. Therefore, more and more in silico models have been constructed for the prediction of the hAOX metabolism. These models are based on molecular docking and quantum chemistry theory, which are time-consuming and difficult to automate. Therefore, in this study, we compared traditional machine learning methods, graph convolutional neural network methods, and sequence-based methods with limited data, and proposed a ligand-based model for the metabolism prediction catalyzed by hAOX. Compared with the published models, our model achieved better performance (ACC = 0.91, F1 = 0.77). What's more, we built a web server to predict the sites of metabolism (SOMs) for hAOX. In summary, this study provides a convenient and automatable model and builds a web server named Meta-hAOX for accelerating the drug design and optimization stage.

CAV2 promotes the invasion and metastasis of head and neck squamous cell carcinomas by regulating S100 proteins.

More than half of HNSCC patients are diagnosed with advanced disease. Locally advanced HNSCC is characterized by tumors with marked local invasion and evidence of metastasis to regional lymph nodes. CAV2 is a major coat protein of caveolins, important components of the plasma membrane. In this study, CAV2 was found to profoundly promote invasion and stimulate metastasis in vivo and in vitro. CAV2 was demonstrated to be a key regulator of S100 protein expression that upregulates the proteins levels of S100s, which promotes the invasion and migration and downregulates the expression of tumor suppressors. Mechanistically, CAV2 directly interacts with S100s in HNSCC cells, and CAV2 reduces S100A14 protein expression by promoting its ubiquitylation and subsequent degradation via the proteasome. Moreover, we discovered that CAV2 promotes the interaction between S100A14 and the E3 ubiquitin ligase TRIM29 and increases TRIM29 expression. Taken together, our findings indicate that CAV2 promotes HNSCC invasion and metastasis by regulating the expression of S100 proteins, presenting a novel potential target for anticancer therapy in HNSCC.

The functions and prognostic value of Krüppel-like factors in breast cancer.

BACKGROUND: Krüppel-like factors (KLFs) are zinc finger proteins which participate in transcriptional gene regulation. Although increasing evidence indicate that KLFs are involved in carcinogenesis and progression, its clinical significance and biological function in breast cancer are still limited. METHODS: We investigated all the expression of KLFs (KLF1-18) at transcriptional levels by using Oncomine and Gene Expression Profiling Interactive Analysis (GEPIA). The mRNA and protein expression levels of KLFs were also determined by using RT-qPCR and immunohistochemistry, respectively. CBioPortal, GeneMANIA and STRING were used to comprehensive analysis of the molecular characteristics of KLFs. The clinical value of prognostic prediction based on the expression of KLFs was determined by using the KM plotter. The relevant molecular pathways of KLFs were further analyzed by using Gene Set Enrichment Analysis (GSEA) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database. Finally, we investigated the effect of KLF2 and KLF15 on biological behavior of breast cancer cells in vitro. RESULTS: The expression of KLF2/4/6/8/9/11/15 was significantly down-regulated in breast cancer. The patients with high KLF2, KLF4 or KLF15 expression had a better outcome, while patients with high KLF8 or KLF11 had a poor prognosis. Furthermore, our results showed that KLF2 or KLF15 can be used as a prognostic factor independent on the other KLFs in patients with breast cancer. Overexpression of KLF2 or KLF15 inhibited cell proliferation and migration, and blocked cell cycle at G0/G1 phase, resulting in cell apoptosis. CONCLUSIONS: KLF2 and KLF15 function as tumor suppressors in breast cancer and are potential biomarkers for prognostic prediction in patients with breast cancer.

The Neutrophil-to-Lymphocyte Ratio (NLR) Predicts the Prognosis of Unresectable Intermediate and Advanced Hepatocellular Carcinoma Treated with Apatinib.

PURPOSE: Patients with hepatocellular carcinoma (HCC) who might benefit most from anti-angiogenesis therapy remain unknown. In recent years, neutrophil-to-lymphocyte ratio (NLR), an indicator of inflammatory response, has received particular attention in HCC. Herein, we explored the prognostic value of pre-treatment NLR in individuals with unresectable intermediate and advanced hepatocellular carcinoma treated with apatinib, a second-line angiogenesis inhibitor. The findings of this study would assist in precision medicine and provide clinical decision support. PATIENTS AND METHODS: This is a retrospective study in which 171 HCC patients attending Tianjin Medical University Cancer Institute and Hospital and treated with apatinib between January 2016 and July 2018 were enrolled. The prognosis of the patients based on NLR signatures was then analyzed. RESULTS: Patients with a low pre-treatment NLR (NLR < 2.49) presented a significantly longer overall survival (OS) (P < 0.001) and progression-free survival (PFS) (P = 0.043). Furthermore, a low pre-treatment NLR level could be used to predict a longer OS in patients with non-macrovascular invasion (P < 0.001). Independent of serum alpha-fetoprotein (AFP) levels, a low NLR level in this cohort of patients is associated with a longer OS. CONCLUSION: Pre-treatment NLR predicts the prognosis of patients with unresectable intermediate and advanced HCC treated with apatinib.

LDLR inhibition promotes hepatocellular carcinoma proliferation and metastasis by elevating intracellular cholesterol synthesis through the MEK/ERK signaling pathway.

OBJECTIVE: Adaptive rewiring of cancer energy metabolism has received increasing attention. By binding with LDLs, LDLRs make most of the circulating cholesterol available for cells to utilize. However, it remains unclear how LDLR works in HCC development by affecting cholesterol metabolism. METHODS: Database analyses and immunohistochemical staining were used to identify the clinical significance of LDLR in HCC. A transcriptome analysis was used to reveal the mechanism of LDLR aberration in HCC progression. A liver orthotopic transplantation model was used to evaluate the role of LDLR in HCC progression in vivo. RESULTS: Downregulation of LDLR was identified as a negative prognostic factor in human HCC. Reduced expression of LDLR in HCC cell lines impaired LDL uptake but promoted proliferation and metastasis in vitro and in vivo. Mechanistically, increasing intracellular de novo cholesterol biosynthesis was the chief contributor to malignant behaviors caused by LDLR inhibition, which could be rescued by simvastatin. Activation of the MEK/ERK pathway by LDLR downregulation partially contributed to intracellular cholesterol synthesis in HCC. CONCLUSIONS: Downregulation of LDLR may elevate intracellular cholesterol synthesis to accelerate proliferation and motility through a mechanism partially attributed to stimulation of the MEK/ERK signaling pathway. Repression of intracellular cholesterol synthesis with statins may constitute a targetable liability in the context of lower LDLR expression in HCC.

Angiotensin-converting enzyme inhibitors have adverse effects in anti-angiogenesis therapy for hepatocellular carcinoma.

At present, anti-angiogenic drugs (AADs) are widely used in the systemic treatment of hepatocellular carcinoma (HCC) or other types of cancer, and have achieved good anti-cancer effect, whereas treatment-related proteinuria can affect the routine use of AADs, which in turn abates the overall efficacy. Currently, most clinicians prescribe angiotensin-converting enzyme inhibitors (ACEIs) to alleviate proteinuria according to diabetic nephropathy guidelines or expert recommendations. However, the efficacy of ACEIs in reducing AAD-related proteinuria and its effect on the anticancer effect of AADs is unknown. Our clinical data showed that some HCC patients experienced tumor progression by ACEIs administration for the treatment of proteinuria caused by AADs. Here, we confirmed that in different tumor-bearing mouse models, ACEIs did not delay the appearance of proteinuria or alleviate proteinuria caused by AADs but compromised the anticancer efficacy of AADs. This effect is unrelated to the change in the VEGF signaling pathway. Our data showed that the combination of ACEIs and AADs flared the production of kidney-derived erythropoietin (EPO). In turn, EPO compromises the anti-angiogenic effects of AADs and decreases antitumor activity. In conclusion, for the treatment of proteinuria caused by AADs, ACEIs have no efficacy while also promoting AADs resistance. This finding is of great significance to guide clinical standardized management of side effects of anti-angiogenic therapy for cancer patients.

Apatinib as first-line treatment in patients with advanced hepatocellular carcinoma: a phase II clinical trial.

BACKGROUND: The prognosis for advanced hepatocellular carcinoma (HCC) remains clinically unsatisfying. Apatinib has proven to be a very effective treatment for advanced HCC in our previous retrospective study. Our aim in this study was to evaluate the efficacy, safety, and toxicity of apatinib in patients with advanced HCC. METHODS: This single-arm, open-label phase II clinical trial enrolled patients with advanced HCC. These patients received apatinib, 500 mg once daily, until disease progression, unacceptable toxicity, consent withdrawal, or death. One treatment cycle consisted of 4 weeks of apatinib treatment. The response evaluation criteria in solid tumors (RECIST) was used to assess tumor response every 1-2 cycles. The primary endpoint was the objective response rate (ORR), while the secondary endpoints were the overall survival (OS), progression-free survival (PFS), disease control rate (DCR), and toxicity. RESULTS: Between December 2016 and June 2018, 23 patients were enrolled in the study, 22 of whom were available for response evaluation. The cutoff date was August 10, 2018. The overall ORR and DCR were 30.4% and 65.2%, respectively. The median OS and PFS were 13.8 (95% CI: 5.3-22.3) and 8.7 (95% CI: 5.9-11.1) months, respectively. The most common treatment-related adverse events were proteinuria (39.1%), hypertension (34.8%), and hand-foot-skin reaction (34.8%). CONCLUSIONS: Apatinib showed robust clinical activity in patients with advanced HCC. Moreover, apatinib was safe to use, well tolerated, and had acceptable toxicity. (NCT03046979).

Drug-Related Hypertension Associated with the Efficacy of Apatinib on Hepatocellular Carcinoma.

PURPOSE: We retrospectively evaluated the efficacy and safety of apatinib as a first-line treatment for advanced hepatocellular carcinoma (HCC) and explored whether drug-related hypertension (HTN) could predict its efficacy. PATIENTS AND METHODS: This retrospective analysis included patients with advanced HCC who received oral treatment with apatinib. We evaluated the effectiveness by overall survival (OS), progression-free survival (PFS), time to progression (TTP), and disease control rate (DCR), and assessed the safety of the drug based on the occurrence of adverse events. In order to explore whether apatinib-related HTN can be used as a predictor of therapeutic effect, patients were divided into an HTN group and a non-HTN group and adjusted for propensity score-matched (PSM) to reduce mixed deviation. Subgroup analyses of negative prognostic factors for advanced HCC were also performed, including alpha-fetoprotein (AFP), Child-Pugh Score, macrovascular invasion, and extrahepatic metastasis. RESULTS: A total of 208 patients were analyzed, of which 40.9% (n =85) developed drug-related HTN. For all patients, the OS was 13.4 months (95% CI, 12.2-14.6), the PFS was 5.7 months (95% CI, 5.1-6.3), and the TTP was 6.9 months (95% CI, 6.0-7.8). The OS of the HTN group and the non-HTN group was 17.4 months (m) and 12.5m (p=0.001), and the PFS was 7.4m and 4.7m (p=0.000), respectively. After PSM, the OS (p=0.001) and PFS (p=0.003) of the HTN group were still significantly better than the non-HTN group. Subgroup analysis suggested that overall survival was significantly longer in patients with HTN when serum AFP ≤400 µg/L or extrahepatic metastases. Moreover, OS in the HTN group increased significantly with or without macrovascular invasion. In addition, through the analysis of two groups of patients with PFS>6m and PFS≤6m, we know that the patients with drug-related HTN may develop resistance later, so they have longer survival time. CONCLUSION: Apatinib demonstrates compelling anti-cancer activity and acceptable safety in advanced HCC. Apatinib-related HTN can potentially predict prolonged survival in patients with advanced HCC.

Resection of "down-staged" advanced hepatocellular carcinoma after treatment with the VEGFR2 inhibitor apatinib: five cases report.

Sorafenib and lenvatinib are currently standard treatments for advanced hepatocellular carcinoma (HCC); however, the therapeutic effect is unsatisfying. Indeed, very few patients with HCC under sorafenib treatment were eligible for surgery in the past ten years. In addition, there is no report of a patient with the opportunity to undergo radical resection after treatment with lenvatinib. Here, we describe five patients with advanced and unresectable HCC that were able to receive curative resection within 1 year of treatment with the tyrosine kinase inhibitor apatinib that selectively inhibits vascular endothelial growth factor receptor 2 (VEGFR2). The five patients with advanced and unresectable HCC were treated with apatinib (250 mg po, qd), and all the five patients obtained an objective response to the treatment, allowing for subsequent resection, and the second patient even obtained a pathological complete response. The latest follow-up date was August 20, 2019, and all patients were alive at the latest follow-up. The disease-free survival of the first patient was 13 months. Lung metastasis was found 12 months later after surgery for patient 5. The other three patients have no recurrence. This is the first report of a single drug with promising therapeutic effects in patients with advanced HCC within one year at a single center. Therefore, apatinib may be promising for some patients with locally advanced HCC to undergo radical resection and improve outcomes.