Hsc70 promotes anti-tumor immunity by targeting PD-L1 for lysosomal degradation.

Immune checkpoint inhibition targeting the PD-1/PD-L1 pathway has become a powerful clinical strategy for treating cancer, but its efficacy is complicated by various resistance mechanisms. One of the reasons for the resistance is the internalization and recycling of PD-L1 itself upon antibody binding. The inhibition of lysosome-mediated degradation of PD-L1 is critical for preserving the amount of PD-L1 recycling back to the cell membrane. In this study, we find that Hsc70 promotes PD-L1 degradation through the endosome-lysosome pathway and reduces PD-L1 recycling to the cell membrane. This effect is dependent on Hsc70-PD-L1 binding which inhibits the CMTM6-PD-L1 interaction. We further identify an Hsp90α/ß inhibitor, AUY-922, which induces Hsc70 expression and PD-L1 lysosomal degradation. Either Hsc70 overexpression or AUY-922 treatment can reduce PD-L1 expression, inhibit tumor growth and promote anti-tumor immunity in female mice; AUY-922 can further enhance the anti-tumor efficacy of anti-PD-L1 and anti-CTLA4 treatment. Our study elucidates a molecular mechanism of Hsc70-mediated PD-L1 lysosomal degradation and provides a target and therapeutic strategies for tumor immunotherapy.

Application of artificial intelligence radiomics in the diagnosis, treatment, and prognosis of hepatocellular carcinoma.

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer, with an increasing incidence and poor prognosis. In the past decade, artificial intelligence (AI) technology has undergone rapid development in the field of clinical medicine, bringing the advantages of efficient data processing and accurate model construction. Promisingly, AI-based radiomics has played an increasingly important role in the clinical decision-making of HCC patients, providing new technical guarantees for prediction, diagnosis, and prognostication. In this review, we evaluated the current landscape of AI radiomics in the management of HCC, including its diagnosis, individual treatment, and survival prognosis. Furthermore, we discussed remaining challenges and future perspectives regarding the application of AI radiomics in HCC.

Impact of sarcopenia on the short-term and long-term outcomes of intrahepatic cholangiocarcinoma undergoing hepatectomy: A multi-center study.

BACKGROUND: Sarcopenia is associated with adverse prognosis of intrahepatic cholangiocarcinoma (iCCA) after surgery. METHODS: 321 patients with iCCA undergoing surgery were retrospectively recruited and assigned to training and validation cohort. Skeletal muscle index (SMI) was assessed to define sarcopenia. Logistic regression and cox regression analysis were used to identify risk factors. A novel sarcopenia-based nomogram was constructed and validated by ROC curves, calibration curves, and DCA curves. RESULTS: 260 patients were included for analysis. The median age was 63.0 years and 161 patients (61.9%) were diagnosed with sarcopenia. Patients with sarcopenia exhibited a higher rate of postoperative complications, a worse OS and RFS than patients without sarcopenia. Sarcopenia, low albumin and intraoperative blood transfusion were independent risk factors of postoperative complications, while sarcopenia and low albumin were risk factors of high CCI≥26.2. Sarcopenia, high PS score, low-undifferentiated differentiation, perineural invasion, TNM stage III-IV were risk factors of OS, and a novel nomogram based on these five factors was built to predict the 12-, 24-, and 36-months OS, with the mean AUC > 0.6. CONCLUSION: Sarcopenia is negatively associated with both postoperative complications and survival prognosis of iCCA undergoing hepatectomy.

Low-Frequency Ultrasound Sensitive Piezo1 Channels Regulate Keloid-Related Characteristics of Fibroblasts.

Keloids are benign fibroproliferative tumors that severely diminish the quality of life due to discomfort, dysfunction, and disfigurement. Recently, ultrasound technology as a noninvasive adjuvant therapy is developed to optimize treatment protocols. However, the biophysical mechanisms have not yet been fully elucidated. Here, it is proposed that piezo-type mechanosensitive ion channel component 1 (Piezo1) plays an important role in low-frequency sonophoresis (LFS) induced mechanical transduction pathways that trigger downstream cellular signaling processes. It is demonstrated that patient-derived primary keloid fibroblasts (PKF), NIH 3T3, and HFF-1 cell migration are inhibited, and PKF apoptosis is significantly increased by LFS stimulation. And the effects of LFS is diminished by the application of GsMTx-4, the selective inhibitor of Piezo1, and the knockdown of Piezo1. More importantly, the effects of LFS can be imitated by Yoda1, an agonist of Piezo1 channels. Establishing a patient-derived xenograft keloid implantation mouse model further verified these results, as LFS significantly decreased the volume and weight of the keloids. Moreover, blocking the Piezo1 channel impaired the effectiveness of LFS treatment. These results suggest that LFS inhibits the malignant characteristics of keloids by activating the Piezo1 channel, thus providing a theoretical basis for improving the clinical treatment of keloids.

Mimicking Tumor Metastasis Using a Transwell-Integrated Organoids-On-a-Chip Platform.

The mortality rate among cancer patients is primarily attributed to tumor metastasis. The evaluation of metastasis potential provides a powerful framework for personalized therapies. However, little work has so far been undertaken to precisely model tumor metastasis in vitro, hindering the development of preventive and therapeutic interventions. In this work, a tumor-metastasis-mimicked Transwell-integrated organoids-on-a-chip platform (TOP) for precisely evaluating tumor metastatic potential is developed. Unlike the conventional Transwell device for detecting cell migration, the engineered device facilitates the assessment of metastasis in patient-derived organoids (PDO). Furthermore, a novel Transwell chamber with a hexagon-shaped structure is developed to mimic the migration of tumor cells into surrounding tissues, allowing for the evaluation of tumor metastasis in a horizontal direction. As a proof-of-concept demonstration, tumor organoids and metastatic clusters are further evaluated at the protein, genetic, and phenotypic levels. In addition, preliminary drug screening is undertaken to highlight the potential for using the device to combat cancers. In summary, the tumor-metastasis-mimicked TOP offers unique capabilities for evaluating the metastasis potential of tumor organoids and contributes to the development of personalized cancer therapies.

Characterizing perceived usability and its correlation with smoking cessation: An analysis of user assessments of the smoking cessation app quitSTART.

Background: As smartphone ownership has become common in all demographic groups in the United States, smartphone applications (apps) for smoking cessation have grown in popularity due to their potential for supporting the diverse populations in the United States who are attempting to quit smoking. Usability is commonly assessed in mobile health (mHealth) technology as an important aspect of the user experience that could influence users' adherence to a health app and health outcomes. However, the variation of perceived usability across demographic groups, and the implications of that variation for app success, have not been well studied. Objective: The aims of this study were to characterize variation in the perceived usability of the National Cancer Institute Smokefree.gov Initiative smoking cessation app quitSTART across demographic groups, and to assess the correlation between perceived usability and short-term smoking cessation. Methods: We conducted a secondary analysis of data from a randomized controlled trial conducted from 2020 to 2021, which used a 16-item modified version of the mHealth App Usability Questionnaire (MAUQ) to quantify perceived usability four weeks after app download among 131 smokers attempting to quit. Responses were coded on a 5-point Likert-type scale ranging from strongly disagree (1) to strongly agree (5) and total perceived usability was calculated as the sum of all 16 items (range: 16-80). Associations between participant demographic characteristics (gender, race, education level, age, etc.) and total usability were determined using an ANCOVA model. A multivariable logistic regression model was used to assess the association between usability and smoking cessation, also assessed 4 weeks after app download. Results: The ANCOVA model demonstrated that race was associated with perceived usability, with participants from a racial minority group reporting higher total usability than White participants (p < 0.001). White participants had an adjusted mean total usability of 55.8 (95 % CI: 52.8, 58.8) while racial minority group participants had an adjusted mean total usability of 66.5 (95 % CI: 61.2, 71.6). Other participant demographic characteristics, such as gender and sexual minority status, were not associated with mean total usability. Total usability was positively associated with smoking cessation (OR: 1.04, 95 % CI: 1.00, 1.08, p = 0.031). Conclusions: Total perceived usability of quitSTART was higher among adults from a racial minority group compared to White adults, and perceived usability was positively associated with cessation success. These findings emphasize the importance of ensuring high usability of mHealth smoking cessation apps for diverse populations.

Crosstalk between endothelial progenitor cells and HCC through periostin/CCL2/CD36 supports formation of the pro-metastatic microenvironment in HCC.

Metastasis causes most cancer-related deaths, and the role and mechanism of periostin (POSTN) in the metastasis of hepatocellular carcinoma (HCC) remain undiscovered. In this study, DEN and HTVi HCC models were performed in hepatic-specific Postn ablation and Postn knock-in mouse to reveal the role of POSTN in HCC metastasis. Furthermore, POSTN was positively correlated with circulating EPCs level and promoted EPC mobilization and tumour infiltration. POSTN also mediated the crosstalk between HCC and EPCs, which promoted metastasis ability and upregulated CD36 expression in HCC through indirect crosstalk. Chemokine arrays further revealed that hepatic-derived POSTN induced elevated CCL2 expression and secretion in EPCs, and CCL2 promoted prometastatic traits in HCC. Mechanistic studies showed that POSTN upregulated CCL2 expression in EPCs via the αvß3/ILK/NF-κB pathway. CCL2 further induced CD36 expression via the CCR2/STAT3 pathway by directly binding to the promoter region of CD36. Finally, CD36 was verified to have a prometastatic role in vitro and to be correlated with POSTN expression, metastasis and recurrence in HCC in clinical samples. Our findings revealed that crosstalk between HCC and EPCs is mediated by periostin/CCL2/CD36 signalling which promotes HCC metastasis and emphasizes a potential therapeutic strategy for preventing HCC metastasis.

Sleep Quality in Patients With Ocular Graft-Versus-Host Disease After Allogeneic Hematopoietic Stem Cell Transplantation.

OBJECTIVES: To investigate the sleep quality in patients with ocular graft-versus-host disease (oGVHD) compared with patients without oGVHD after allogeneic hematopoietic stem cell transplantation (alloHCT) and healthy controls. METHODS: This cross-sectional study analyzed 142 patients after alloHCT including 94 patients with oGVHD and 48 without. Fifty healthy controls were also enrolled. oGVHD was diagnosed according to International Chronic Ocular GVHD Consensus Group (ICOGCG) criteria. Sleep quality was assessed by the Chinese version of the Pittsburgh Sleep Quality Index (CPSQI). Poor sleep quality was defined as CPQSI score greater than 6. RESULTS: Patients after alloHCT demonstrated a significantly higher CPQSI score than those of controls {7.0 [interquartile range (IQR) 5.0-10.0] vs. 5.5 [IQR 4.8-7.0], P =0.002}, especially in the oGVHD subgroup (7.5 [IQR 5.0-11.0] vs. 6.0 [IQR 5.0-8.0], P =0.04) with nearly double prevalence of poor sleep quality (58 [62%] vs. 18 [37%], P =0.006). Poor sleep quality was strikingly correlated with oGVHD diagnosis (adjusted odds ratio [OR]=2.55, 95% confidence interval [CI]: 1.02-6.34, P =0.04) and systemic immunosuppressants (adjusted OR=2.61, 95% CI: 1.32-5.71, P =0.02). Among the ocular parameters, poor sleep quality was significantly associated with higher ICOGCG score (adjusted OR=1.20, 95% CI: 1.03-1.39, P =0.02) and lower tear film break-up time (adjusted OR=0.85, 95% CI: 0.74-0.99, P =0.05). CONCLUSIONS: Poor sleep quality was associated with an increased severity of oGVHD and tear film instability in the long-term alloHCT survivorship.

Remote teaching system for robotic surgery and its validation: results of a randomized controlled study.

BACKGROUND: Currently, only a limited number of remote assistance modalities are utilized in the basic phase of robotic surgery training to facilitate the rapid acquisition of robotic surgery skills by surgeons. This study aimed to investigate the benefits of real-time remote surgical robotic skill training based on a multi-channel video recording and playback system. METHODS: We randomly divided 40 medical students without prior expertise in the use of surgical robots into two groups to assess the performance of trainees on a robotic simulator (Mimic dV-Trainer). The remote group received remote training, while the control group received live one-on-one guidance. We compared the learning curves of the two groups based on simulator scores. Furthermore, the NASA task load index (NASA-TLX) scale was used to measure the fatigue load of the trainers. RESULTS: We observed no significant differences in the demographics or initial baseline skill levels between the two groups. Participants in the remote group achieved higher total scores in the Match Board 2 and Thread the Rings 1 exercises compared to the control group. In addition, trainers in the remote group reported lower subjective fatigue load than in the control group. CONCLUSIONS: The remote approach to surgical robotics skills training based on the Remote Teaching System for Robotic Surgery (ReTeRoS) is both feasible and has the potential for large-scale training.

AZGP1 activation by lenvatinib suppresses intrahepatic cholangiocarcinoma epithelial-mesenchymal transition through the TGF-ß1/Smad3 pathway.

Intrahepatic cholangiocarcinoma (ICC) is a primary liver malignancy and is characterized by highly aggressive and malignant biological behavior. Currently, effective treatment strategies are limited. The effect of lenvatinib on ICC is unknown. In this study, we found that AZGP1 was the key target of lenvatinib in ICC, and its low expression in ICC cancer tissues was associated with a poor prognosis in patients. Lenvatinib is a novel AZGP1 agonist candidate for ICC that inhibits ICC-EMT by regulating the TGF-ß1/Smad3 signaling pathway in an AZGP1-dependent manner. Furthermore, we found that lenvatinib could increase AZGP1 expression by increasing the acetylation level of H3K27Ac in the promoter region of the AZGP1 gene, thereby inhibiting EMT in ICC cells. In conclusion, lenvatinib activates AZGP1 by increasing the acetylation level of H3K27Ac on the AZGP1 promoter region and regulates the TGF-ß1/Smad3 signaling pathway in an AZGP1-dependent manner to inhibit ICC-EMT. This study offers new insight into the mechanism of lenvatinib in the treatment of ICC and provides a theoretical basis for new treatment methods.

Obstruction of the Tear Drainage Altered Lacrimal Gland Structure and Function.

Purpose: Orbital glands and drainage conduits are two distinct entities that constitute the lacrimal apparatus system, the malfunction of which leads to a range of ocular surface disorders. Despite the close functional relationship, how the two parts interact under pathophysiological conditions has not been directly tested. The study aims to investigate the lacrimal gland (LG) structural and functional changes upon the drainage system obstruction, thus, testing their function link. Methods: Dacryocystectomy was performed in C57BL/6 mice to create a surgical model for tear duct (TD) obstruction (STDOB). Prickle1 mutant line with congenital nasolacrimal duct dysplasia serves as a genetic model for TD obstruction (GTDOB). Alterations of the LG and the ocular surface in tear duct obstruction mice were examined. Results: STDOB and GTDOB mice showed similar ocular surface phenotypes, including epiphora, corneal epithelial defects, and conjunctival goblet cell abnormalities. At the molecular and cellular levels, aberrant secretory vesicle fusion of the LG acinar cells was observed with altered expression and localization of Rab3d, Vamp8, and Snap23, which function in membrane fusion. LG secretion was also altered in that lactoferrin, lipocalin2, and lysozyme expression were increased in both LG and tears. Furthermore, STDOB and GTDOB mice exhibited similar LG transcription profiles. Conclusions: Physical obstruction of tear drainage in STDOB or GTDOB mice leads to LG dysfunction, suggesting a long-distance interaction between the tear drainage conduits and the LG. We propose that various components of the lacrimal apparatus should be considered an integral unit in diagnosing and treating ocular surface diseases.

ARIH1 activates STING-mediated T-cell activation and sensitizes tumors to immune checkpoint blockade.

Despite advances in cancer treatment, immune checkpoint blockade (ICB) only achieves complete response in some patients, illustrating the need to identify resistance mechanisms. Using an ICB-insensitive tumor model, here we discover cisplatin enhances the anti-tumor effect of PD-L1 blockade and upregulates the expression of Ariadne RBR E3 ubiquitin-protein ligase 1 (ARIH1) in tumors. Arih1 overexpression promotes cytotoxic T cell infiltration, inhibits tumor growth, and potentiates PD-L1 blockade. ARIH1 mediates ubiquitination and degradation of DNA-PKcs to trigger activation of the STING pathway, which is blocked by the phospho-mimetic mutant T68E/S213D of cGAS protein. Using a high-throughput drug screen, we further identify that ACY738, less cytotoxic than cisplatin, effectively upregulates ARIH1 and activates STING signaling, sensitizing tumors to PD-L1 blockade. Our findings delineate a mechanism that tumors mediate ICB resistance through the loss of ARIH1 and ARIH1-DNA-PKcs-STING signaling and indicate that activating ARIH1 is an effective strategy to improve the efficacy of cancer immunotherapy.

Polymorphonuclear myeloid-derived suppressor cells play a proinflammatory role via TNF-α+ B cells through BAFF/BTK/NF-κB signalling pathway in the pathogenesis of collagen-induced arthritis mice.

Although various studies have been performed on the function of polymorphonuclear myeloid-derived suppressor cells (PMN-MDSCs) in RA, the results were conflicting. Here we were trying to clarify the role of PMN-MDSCs in the pathogenesis of RA and its specific mechanisms. We detected the frequencies and counts of PMN-MDSCs, TNF-α+ B cells and Ki67+ B cells in spleen and inflamed joints of collagen-induced arthritis (CIA) mice using flow cytometry. The pathological role of PMN-MDSCs was examined by anti-Ly6G neutralizing antibodies against PMN-MDSCs or adoptive transfer of PMN-MDSCs. And the modulation of PMN-MDSCs on B cells was conducted by coculture assays, RNA-Seq, RT-qPCR, and so on. The mechanism of BAFF regulating B cells was verified through western blot and flow cytometry. PMN-MDSCs accumulated in the spleen and joints of CIA mice. PMN-MDSCs depletion could alleviate the arthritis severity, which was accompanied by decreased TNF-α secretion and proliferation of B cells. And its adoptive transfer also facilitated disease progress. Furthermore, PMN-MDSCs from CIA mice had higher expression level of BAFF, which regulated TNF-α expression, proliferation and apoptosis of B cells in vitro. What's more, BAFF promoted phosphorylation of BTK/NF-κB signalling pathway. And Ibrutinib (BTK inhibitor) could reverse the effect of BAFF on TNF-α expression of B cells. Our study suggested that PMN-MDSCs enhanced disease severity of CIA and manipulated TNF-α expression, proliferation and apoptosis of B cells via BAFF, furthermore, BAFF promoted TNF-α expression through BTK/NF-κB signalling pathway, which demonstrated a novel pathogenesis of PMN-MDSCs in CIA.

Exploring the use of driving simulation to improve robotic surgery simulator training: an observational case-control study.

The correlation between driving skills and the ability to perform robotic surgery have not yet been discussed. Therefore, this study aimed to investigate the impact of driving skills on learning robotic surgery using a driving simulator and a robotic simulator. Sixty robot- and simulator-naïve participants were recruited: 30 with a driver's license and 30 without a driver's license. All participants completed a test on the driving simulator and learned four tasks using a robotic surgical simulator (dV-Trainer). On the driving simulator, the lap time in the driver's license group (D-Group) was significantly lower than that in the non-driver's license group (ND-Group) [217.93 ± 42.79 s vs. 271.24 ± 46.63 s, P < 0.001]. The average number of tires off track in the D-Group was lower than that in the ND-Group (0.13 ± 0.35 vs. 0.57 ± 0.63, P = 0.002). The baseline score of the D-Group on the robotic simulator was higher than that of the ND-Group (467.53 ± 107.62 vs. 385.53 ± 136.30, P = 0.022). In the Pick-and-Place-Clutching, Peg-Board-2, and Thread-the-Rings-1 tasks, the learning curve of the D-Group was steeper than that of the ND-Group. However, no significant difference was observed in the Match-Board-2 task. According to the lap time ranking, participants in the top tertile had a steeper learning curve than those in the bottom tertile, especially for the Pick-and-Place-Clutching and Peg-Board-2 tasks (P < 0.05). Significant differences were also found in the baseline and final stages of the Thread-the-Rings-1 task and in the initial stage of the Match-Board-2 task (P < 0.05). Students with a driver's license or better performance in racing games had more success in learning robotic surgery. Driving simulators may promote robotic surgery training.

Pharmacological Mechanisms and Clinical Applications of Curcumin: Update.

Curcumin, a well-known hydrophobic polyphenol extracted from the rhizomes of turmeric (Curcuma longa L.), has attracted great interest in the last ten years due to its multiple pharmacological activities. A growing body of evidence has manifested that curcumin has extensive pharmacological activities including anti-inflammatory, anti-oxygenation, lipid regulation, antiviral, and anticancer with hypotoxicity and minor adverse reactions. However, the disadvantages of low bioavailability, short half-life in plasma, low drug concentration in blood, and poor oral absorption severely limited the clinical application of curcumin. Pharmaceutical researchers have carried out plenty of dosage form transformations to improve the druggability of curcumin and have achieved remarkable results. Therefore, the objective of this review summarizes the pharmacological research progress, problems in clinical application and the improvement methods of curcumin's druggability. By reviewing the latest research progress of curcumin, we believe that curcumin has a broad clinical application prospect for its wide range of pharmacological activities with few side effects. The deficiencies of lower bioavailability of curcumin could be improved by dosage form transformation. However, curcumin in the clinical application still requires further study regarding the underlying mechanism and clinical trial verification.

A 3D-printed PCL/PEI/DNA bioactive scaffold for chemotherapy drug capture in vivo.

Systemic chemotherapy after surgery is necessary to control tumor recurrence, but the severe side effects caused by chemotherapeutic drugs pose a great threat to patients' health. In this study, we originally develop a porous scaffold used for chemotherapy drug capture by using 3D printing technology. The scaffold is mainly composed of poly (ε-caprolactone) (PCL) and polyetherimide (PEI) with a mass ratio of 5/1. Subsequently, the printed scaffold is modified with DNA through the strong electrostatic integration between DNA and PEI to endow the scaffold with the specific absorption to doxorubicin (DOX, a widely used chemotherapy drug). The results show that pore diameter has an important influence on DOX adsorption, and smaller pores will ensure a higher DOX absorption. In vitro, the printed scaffold can absorb about 45 % DOX. While in vivo, it remains a higher absorption ability to DOX when the scaffold is successfully implanted into the common jugular vein of rabbits. What's more, the scaffold has good hemocompatibility and biocompatibility, indicating its safety for in vivo application. Taken together, the 3D-printed scaffold with excellent capture of chemotherapy drugs will play an important role in reducing the toxic side effects of chemotherapy drugs and improving the life quality of patients.

Interleukin-13 reduces bone erosion in rheumatoid arthritis by up-regulating osteoprotegerin expression in fibroblast-like synoviocytes : an in vitro and in vivo study.

OBJECTIVES: Bone erosion in rheumatoid arthritis (RA) is partly caused by excessive activation of osteoclasts. Osteoclasts can be derived from RA synovium and their differentiation can be inhibited by osteoprotegerin (OPG), a decoy receptor of the osteoclastogenesis-promoting cytokine receptor activator of nuclear factor κB ligand (RANKL). Fibroblast-like synoviocytes (FLSs) are the main stromal cells in the synovium that can secret OPG. The OPG secretion of FLSs can be modulated by various cytokines. Interleukin (IL)-13 can alleviate bone erosion in RA mouse models, but the mechanisms remain unclear. Therefore, we aimed to investigate whether IL-13 can induce OPG secretion by RA-FLSs, thus ameliorating bone destruction in RA by inhibiting osteoclast differentiation. METHODS: OPG, RANKL, and IL-13 receptors expression by RA-FLSs were evaluated by RT-qPCR. OPG secretion was determined by ELISA. Western blot was performed to analyse OPG expression and the activation of the STAT6 pathway. IL-13 and (or) OPG siRNA pre-treated RA-FLSs conditioned medium were used in osteoclast induction to test if IL-13 can inhibit osteoclastogenesis by up-regulating OPG in RA-FLSs. Micro-CT and immunofluorescence were performed to determine if IL-13 can induce OPG expression and alleviate bone erosion in vivo. RESULTS: IL-13 can promote OPG expression of RA-FLSs, and the promotion can be overcome by IL-13Rα1 or IL-13Rα2 siRNA transfection, or STAT6 inhibitor. Osteoclast differentiation can be inhibited by IL-13 pre-treated RA-FLSs conditioned medium. The inhibition can be reversed by OPG siRNA transfection. IL-13 injection can increase OPG expression in the joints while reducing bone destruction in collagen-induced arthritis mice. CONCLUSIONS: IL-13 can inhibit osteoclastogenesis by up-regulating OPG in RA-FLSs through IL-13 receptors via the STAT6 pathway, thus may ameliorate bone erosion in RA.

Prediction of Response to Lenvatinib Monotherapy for Unresectable Hepatocellular Carcinoma by Machine Learning Radiomics: A Multicenter Cohort Study.

PURPOSE: We aimed to construct machine learning (ML) radiomics models to predict response to lenvatinib monotherapy for unresectable hepatocellular carcinoma (HCC). EXPERIMENTAL DESIGN: Patients with HCC receiving lenvatinib monotherapy at three institutions were retrospectively identified and assigned to training and external validation cohorts. Tumor response after initiation of lenvatinib was evaluated. Radiomics features were extracted from contrast-enhanced CT images. The K-means clustering algorithm was used to distinguish radiomics-based subtypes. Ten ML radiomics models were constructed and internally validated by 10-fold cross-validation. These models were subsequently verified in an external validation cohort. RESULTS: A total of 109 patients were identified for analysis, namely, 74 in the training cohort and 35 in the external validation cohort. Thirty-two patients showed partial response, 33 showed stable disease, and 44 showed progressive disease. The overall response rate (ORR) was 29.4%, and the disease control rate was 59.6%. A total of 224 radiomics features were extracted, and 25 significant features were identified for further analysis. Two distant radiomics-based subtypes were identified by K-means clustering, and subtype 1 was associated with a higher ORR and longer progression-free survival (PFS). Among the 10 ML algorithms, AutoGluon displayed the highest predictive performance (AUC = 0.97), which was relatively stable in the validation cohort (AUC = 0.93). Kaplan-Meier analysis showed that responders had a better overall survival [HR = 0.21; 95% confidence interval (CI): 0.12-0.36; P < 0.001] and PFS (HR = 0.14; 95% CI: 0.09-0.22; P < 0.001) than nonresponders. CONCLUSIONS: Valuable ML radiomics models were constructed, with favorable performance in predicting the response to lenvatinib monotherapy for unresectable HCC.