Automated whole slide image analysis for a translational quantification of liver fibrosis

Assessing the diagnostic accuracy of ChatGPT-4 in the histopathological evaluation of liver fibrosis in MASH

BACKGROUND

Large language models like ChatGPT have demonstrated potential in medical image interpretation, but their efficacy in liver histopathological analysis remains largely unexplored. This study aims to assess ChatGPT-4-vision's diagnostic accuracy, compared to liver pathologists' performance, in evaluating liver fibrosis (stage) in metabolic dysfunction-associated steatohepatitis.

METHODS

Digitized Sirius Red-stained images for 59 metabolic dysfunction-associated steatohepatitis tissue biopsy specimens were evaluated by ChatGPT-4 and 4 pathologists using the NASH-CRN staging system. Fields of view at increasing magnification levels, extracted by a senior pathologist or randomly selected, were shown to ChatGPT-4, asking for fibrosis staging. The diagnostic accuracy of ChatGPT-4 was compared with pathologists' evaluations and correlated to the collagen proportionate area for additional insights. All cases were further analyzed by an in-context learning approach, where the model learns from exemplary images provided during prompting.

RESULTS

ChatGPT-4's diagnostic accuracy was 81% when using images selected by a pathologist, while it decreased to 54% with randomly cropped fields of view. By employing an in-context learning approach, the accuracy increased to 88% and 77% for selected and random fields of view, respectively. This method enabled the model to fully and correctly identify the tissue structures characteristic of F4 stages, previously misclassified. The study also highlighted a moderate to strong correlation between ChatGPT-4's fibrosis staging and collagen proportionate area.

CONCLUSIONS

ChatGPT-4 showed remarkable results with a diagnostic accuracy overlapping those of expert liver pathologists. The in-context learning analysis, applied here for the first time to assess fibrosis deposition in metabolic dysfunction-associated steatohepatitis samples, was crucial in accurately identifying the key features of F4 cases, critical for early therapeutic decision-making. These findings suggest the potential for integrating large language models as supportive tools in diagnostic pathology.

Transposon-based oncogene integration in Abcb4(Mdr2)-/- mice recapitulates high susceptibility to cholangiocarcinoma in primary sclerosing cholangitis

BACKGROUND & AIMS

Cholangiocarcinoma (CCA) is a dreaded complication of primary sclerosing cholangitis (PSC) that is difficult to diagnose and associated with high mortality. A lack of animal models of CCA recapitulating the hepatic microenvironment of sclerosing cholangitis has hindered the development of novel treatments. Herein, we sought to develop a mouse model of PSC-associated CCA.

METHODS

Ten-week-old Mdr2-/- mice with congenital PSC-like disease, and healthy wild-type littermates were subjected to either modified retrograde biliary instillation or hydrodynamic tail vein injection of a sleeping beauty transposon-transposase plasmid system with activated AKT (myr-AKT) and Yap (YapS127A) proto-oncogenes (SB AKT/YAP1). The role of TGFβ was interrogated via ALK5 inhibitor (SB-525334) administration. Tumor phenotype, burden and desmoplastic reaction were analyzed histologically and via RNA sequencing.

RESULTS

While SB AKT/YAP1 plasmids administered via retrograde biliary injection caused tumors in Mdr2-/-, only 26.67% (4/15) of these tumors were CCA. Alternatively, hydrodynamic tail vein injection of SB AKT/YAP1 resulted in robust tumorigenesis in all fibrotic Mdr2-/- mice with high CCA burden compared to healthy mice. Tumors phenotypically resembled human CCA, expressed multiple CCA (but not hepatocellular carcinoma) markers, and exhibited a profound desmoplastic reaction. RNA sequencing analysis revealed profound transcriptional changes in CCA evolving in a PSC-like context, with specific alterations in multiple immune pathways. Pharmacological TGFβ inhibition led to enhanced immune cell tumor infiltration, reduced tumor burden and suppressed desmoplastic collagen accumulation compared to placebo.

CONCLUSION

We established a new high-fidelity cholangiocarcinoma model in mice, termed SB CCA.Mdr2-/-, which recapitulates the increased susceptibility to CCA in the setting of biliary injury and fibrosis observed in PSC. Through transcriptomics and pharmacological studies, we show dysregulation of multiple immune pathways and TGFβ signaling as potential drivers of CCA in a PSC-like microenvironment.

IMPACT AND IMPLICATIONS

Animal models for primary sclerosing cholangitis (PSC)-related cholangiocarcinoma (PSC-CCA) are lacking. Thus, we have developed and characterized a new mouse model of PSC-CCA, termed SB CCA.Mdr2-/-, which features reliable tumor induction on a PSC-like background of biliary injury and fibrosis. Global gene expression alterations were identified and standardized tools, including automated whole slide image analysis methodology for tumor burden and feature analysis, were established to enable systematic research into PSC-CCA biology and formal preclinical drug testing.

Artificial intelligence-assisted digital pathology for non-alcoholic steatohepatitis: current status and future directions

The worldwide prevalence of non-alcoholic steatohepatitis (NASH) is increasing, causing a significant medical burden, but no approved therapeutics are currently available. NASH drug development requires histological analysis of liver biopsies by expert pathologists for trial enrolment and efficacy assessment, which can be hindered by multiple issues including sample heterogeneity, inter-reader and intra-reader variability, and ordinal scoring systems. Consequently, there is a high unmet need for accurate, reproducible, quantitative, and automated methods to assist pathologists with histological analysis to improve the precision around treatment and efficacy assessment. Digital pathology (DP) workflows in combination with artificial intelligence (AI) have been established in other areas of medicine and are being actively investigated in NASH to assist pathologists in the evaluation and scoring of NASH histology. DP/AI models can be used to automatically detect, localise, quantify, and score histological parameters and have the potential to reduce the impact of scoring variability in NASH clinical trials. This narrative review provides an overview of DP/AI tools in development for NASH, highlights key regulatory considerations, and discusses how these advances may impact the future of NASH clinical management and drug development. This should be a high priority in the NASH field, particularly to improve the development of safe and effective therapeutics.

Systemic multiomics evaluation of the therapeutic effect of Bacteroides species on liver cirrhosis in male mice

IMPORTANCE

The human gut microbiome mediates bidirectional interaction within the gut-liver axis, while liver diseases, including liver cirrhosis, are very closely related to the state of the gut environment. Thus, improving the health of the gut-liver axis by targeting the intestinal microbiota is a potential therapeutic approach in hepatic diseases. This study examines changes in metabolomics and microbiome composition by treating bacteria derived from the human gut in mice with liver cirrhosis. Interorgan-based multiomics profiling coupled with functional examination demonstrated that the treatment of Bacteroides dorei pertained to protective effects on liver cirrhosis by normalizing the functional, metabolic, and metagenomic environment through the gut-liver axis. The study provides the potential value of a multiomics-based and interorgan-targeted evaluation platform for the comprehensive examination and mechanistic understanding of a wide range of biologics, including gut microbes. Furthermore, the current finding also suggests in-depth future research focusing on the discovery and validation of next-generation probiotics and products (postbiotics).

The Development of Nonalcoholic Fatty Liver Disease and Metabolic Syndromes in Diet-Induced Rodent Models

Dietary macronutrients are essential for metabolic regulation and insulin function. The present study examined the effects of different high-fat diets (HFDs) and high-carbohydrate diets (HCDs) on the development of non-alcoholic fatty liver disease and metabolic syndrome indices in healthy adult male Wistar albino rats. Forty-two rats were distributed into six groups (n = 7), which were fed the following for 22 weeks: (1) a control diet; (2) a high-carbohydrate, low-fat diet (HCD-LFD); (3) high-saturated-fat, low-carbohydrate diet (HSF-LCD); (4) a high-monounsaturated-fat diet (HMUSF); (5) a high medium-chain fat diet (HMCF); and a (6) a high-carbohydrate, high-fiber diet (HCHF). In comparison to the control, the body weight increased in all the groups. The HSF-LCD group showed the highest levels of cholesterol, triglyceride, low-density lipoprotein, hepatic enzyme, insulin resistance, and Homeostatic Model Assessment for Insulin Resistance. A liver histology analysis of the HSF-LCD group showed macrovesicular hepatic steatosis associated with large hepatic vacuolation. Additionally, it showed marked periportal fibrosis, especially around the blood vessels and blood capillaries. The lowest levels of fasting glycemia, insulin, and HOMA-IR were observed in the HCHF group. In conclusion, these findings show that dietary saturated fat and cholesterol are principal components in the development and progression of non-alcoholic fatty liver disease in rats, while fiber showed the greatest improvement in glycemic control.

Phosphotungstic Acid-treated Picrosirius Red Staining Improves Whole-slide Quantitative Analysis of Collagen in Histological Specimens

We tried to prevent nonspecific nuclear staining (NS-NS) of picrosirius red (PSR) staining by treating the specimens with one of the heteropoly acids phosphotungstic acid (PTA). We analyzed a total of 35 cases of non-cancerous liver tissue for fibrosis and NS-NS under PSR-alone, phosphomolybdic acid (PMA)-pretreated PSR (PMA + PSR), or PTA-pretreated PSR (PTA + PSR) condition. In addition, we analyzed the photosensitivity of PMA or PTA single stain specimens. PTA + PSR significantly suppressed NS-NS compared with PSR. The color of the specimens did not change into blue by 30 times the exposure to whole slide scanner (WSS) light. The PTA + PSR condition showed the highest correlation with the Ishak score (pathological evaluation of liver fibrosis) compared with other conditions. Furthermore, Sirius Red-positive percentage (SRP%) in PSR was increased in the NS-NS observed cases. SRP% in PMA + PSR was significantly affected by WSS light exposure time. Moreover, the deposition of non-polarized PSR-stained substances (NP-PSR+S) clinging to the collagen fibers potentially explains why SRP% seemed bigger under PSR than PTA + PSR. Our protocol enabled us to analyze the whole slide image of PSR staining by high magnification, which would contribute to the accurate analysis of collagen amount in the tissue sections.