Liver metastasis is the leading cause of mortality in patients with colorectal cancer. Given the significance of both epithelial–mesenchymal transition (EMT) of tumor cells and the immune microenvironment in colorectal cancer liver metastasis (CRLM), the interplay between them could hold the key for developing improved treatment options. We employed multiomics analysis of 130 samples from 18 patients with synchronous CRLM integrated with external datasets to comprehensively evaluate the interaction between immune cells and EMT of tumor cells in liver metastasis. Single-cell RNA sequencing analysis revealed distinct distributions of nonmalignant cells between primary tumors from patients with metastatic colorectal cancer (mCRC) and non-metastatic colorectal cancer, showing that Th17 cells were predominantly enriched in the primary lesion of mCRC. TWEAK, a cytokine secreted by Th17 cells, promoted EMT by binding to receptor Fn14 on tumor cells, and the TWEAK–Fn14 interaction enhanced tumor migration and invasion. In mouse models, targeting Fn14 using CRISPR-induced knockout or lipid nanoparticle–encapsulated siRNA alleviated metastasis and prolonged survival. Mice lacking Il17a or Tnfsf12 (encoding TWEAK) exhibited fewer metastases compared with wild-type mice, while cotransfer of Th17 with tumor cells promoted liver metastasis. Higher TWEAK expression was associated with a worse prognosis in patients with colorectal cancer. In addition, CD163L1+ macrophages interacted with Th17 cells, recruiting Th17 via the CCL4-CCR5 axis. Collectively, this study unveils the role of immune cells in the EMT process and identifies TWEAK secreted by Th17 as a driver of CRLM. Significance: TWEAK secreted by Th17 cells promotes EMT by binding to Fn14 on colorectal cancer cells, suggesting that blocking the TWEAK–Fn14 interaction may be a promising therapeutic approach to inhibit liver metastasis.

Background: Disseminating disease knowledge through concise videos on various platforms is an innovative and efficient approach. However, it remains uncertain whether pancreatic neuroendocrine tumor (pNET)-related videos available on current short video platforms can effectively convey accurate and impactful information to the general public. Objective: Our study aims to extensively analyze the quality of pNET-related videos on TikTok and Bilibili, intending to enhance the development of pNET-related social media content to provide the general public with more comprehensive and suitable avenues for accessing pNET-related information. Methods: A total of 168 qualifying videos pertaining to pNETs were evaluated from the video-sharing platforms Bilibili and TikTok. Initially, the fundamental information conveyed in the videos was documented. Subsequently, we discerned the source and content type of each video. Following that, the Global Quality Scale (GQS) and modified DISCERN (mDISCERN) scale were employed to appraise the educational value and quality of each video. A comparative evaluation was conducted on the videos obtained from these two platforms. Results: The number of pNET-related videos saw a significant increase since 2020, with 9 videos in 2020, 19 videos in 2021, 29 videos in 2022, and 106 videos in 2023. There were no significant improvements in the mean GQS or mDISCERN scores from 2020 to 2023, which were 3.22 and 3.00 in 2020, 3.33 and 2.94 in 2021, 2.83 and 2.79 in 2022, and 2.78 and 2.94 in 2023, respectively. The average quality scores of the videos on Bilibili and Tiktok were comparable, with GQS and mDISCERN scores of 2.98 on Bilibili versus 2.77 on TikTok and 2.82 on Bilibili versus 3.05 on TikTok, respectively. The source and format of the videos remained independent factors affecting the two quality scores. Videos that were uploaded by professionals (hazard ratio=7.02, P=.002) and recorded in specialized popular science formats (hazard ratio=12.45, P<.001) tended to exhibit superior quality. Conclusions: This study demonstrates that the number of short videos on pNETs has increased in recent years, but video quality has not improved significantly. This comprehensive analysis shows that the source and format of videos are independent factors affecting video quality, which provides potential measures for improving the quality of short videos.

Table S1. Clinical and pathological characteristics of CRLM patients

Figure S6. Pseudotime analysis of tumor cells by Palantir and CellRank.

Table S2. CD45+ and CD45- cell numbers of each sample from CRLM patients