Introduction

The prevalence of celiac disease is increasing and represents a growing global health issue. The disease profoundly affects patients' quality of life through the multitude of symptoms and complications, as well as the necessity for strict adherence to a gluten-free diet. Gastroenterologists face persistent challenges in diagnosing and managing celiac disease amidst its complexities.
In this context, the integration of artificial intelligence (AI) emerges as a promising advancement, offering potential to revolutionize the landscape of celiac disease management.
Our objective is to evaluate the effectiveness of AI as a reliable and useful asset for patients and gastroenterologists in the management of celiac disease.

Aims & Methods

Research was carried out on the Google Trends website to identify the 5 most searched-for keywords (from 2004 to March 2024) related to celiac disease: gluten-free diet, diagnosis, serologies, associated diseases, complications.

We asked 10 gastroenterologists and 10 celiac patients to formulate all the questions they had about celiac disease in relation to these key words.
Eighty questions were generated based on their responses (40 questions formulated by each group).

The questions were inputted into the "prompt" section of an artificial intelligence (Google GEMINI version 20/02/2024) and the answers were evaluated using a 7-point Likert-type reliability and usefulness scale by two independent celiac disease experts.
Statistical analyses were performed using JAMOVI version 2.3 software. Inter-observer agreement was assessed using Intra-class correlation coefficient [ICC], and Student t-test was used to compare differences between two groups. The significance level for this study was set at p < 0.05.

Results

The intra-class correlation coefficient values for both reliability and usefulness scores demonstrated excellent agreement between the two experts.
Indeed, for answers intended to help gastroenterologists, we observe (Reliability: ICC=0.918, 95% CI : 0.852-0.956 Usefulness: ICC=0.887, 95% CI : 0.798-0.939).
For answers intended for celiac patients (Reliability: ICC=0.935, 95% CI : 0.881-0.965 and ICC=0.926, 95% CI : 0.865-0.960)
The average scores given by experts n°1 and n°2 were: Reliability 4.83 ± 1.58 vs. 5.03 ± 1.54 (p=0.04) and Usefulness 4.75 ± 1.39 vs. 5.00 ± 1.26 (p=0.01), suggesting that AI had limitations in providing reliable but also useful answers to healthcare professionals about celiac disease.
However, when it came to answering celiac patients' questions, we noted that AI gave better scores, this time with no statistically significant difference between the two experts: 5.60 ± 1.15 vs 5.63 ± 1.17 (p=0.78) and 5.67 ± 1.33 vs 5.78 ± 1.27 (p=0.23).
Finally, there was a statistically significant difference between the two groups (gastroenterologist and celiac patients), concerning the reliability and usefulness of AI responses for expert n°1 (p=0.008 and p=.001 respectively) as well as for expert n°2 (p=0.045 and p=0.003 respectively).

Conclusion

The results suggest that AI, particularly Google GEMINI, can be a particularly useful tool in managing celiac disease, especially for celiac patients, by providing reliable and useful answers to questions related to gluten-free diet, diagnosis, serologies, associated diseases, and complications. However, the results are less satisfactory when it comes to answering healthcare professionals, highlighting the need for constant updates and reliance on reliable and up-to-date sources.

Introduction

Aims & Methods

Results

Conclusion

References

Introduction

Intraductal papillary mucinous neoplasm (IPMN) is a significant precancerous lesions of pancreatic cancer, with GNAS mutations frequently observed. We previously demonstrated that GNAS mutations suppress the invasive potential of IPMN-derived cancer cells by inhibition the Notch signaling pathway in a PKA-dependent manner. However, the influence of GNAS mutations on coexisting pancreatic cancers remains unclear. MicroRNAs (miRNAs) are known to regulate gene expression post transcriptionally and can be transferred via extracellular vesicles (EVs) under tumor microenvironmental conditions. In this study, we investigated the paracrine effects of EVs derived from GNAS-mutant IPMN cells and the roles of miRNA cargo in the development and progression of IPMN-associated pancreatic cancers.

Aims & Methods

This study aimed to evaluate the tumor suppressive effects of EVs derived from GNAS-mutant IPMN cells on pancreatic cancer and to identify EV-encapsulated miRNAs that regulate tumor progression. EV-containing supernatants were collected from human pancreatic cancer cells harboring GNAS R201H mutation (established from IPMN-derived invasive cancer) and from a genome-edited GNAS wild-type subline. miRNA profiles in EVs were assessed using GeneChip miRNA microarray. Gene expression was analyzed by qPCR, and protein expression were evaluated via Western blotting.

Results

When EV-containing supernatants from mutant or wild-type GNAS cells were added to pancreatic cancer cells (e.g., Panc-1), those from GNAS-mutant cells significantly suppressed cancer cell invasion and migration. Microarray analysis revealed that miR-602 was more than two-fold upregulated in EVs from mutant cells compared to wild-type cells. Treatment with a PKA inhibitor decreased miR-602 expression in GNAS-mutant cells, whereas PKA activator increased its expression in wild-type cells, indicating that GNAS mutation upregulates miR-602 expression in a PKA signaling. Overexpression of miR-602 in pancreatic cancer cells suppressed their invasive and migratory capabilities. Target gene analysis identified RAC2 and PAK4 –downstream effectors of KRAS pathway– as direct targets of miR-602. These genes are also implicated in PI3K phosphorylation. miR-602 overexpression suppressed PI3K phosphorylation and upregulated E-cadherin expression. These findings suggest that miR-602 act as a tumor suppressor by targeting RAC2/PAK4 and attenuating epithelial-mesenchymal transition (EMT) via PI3K signaling suppression.

Conclusion

IPMNs harboring GNAS mutation upregulate miR-602 expression in EVs in a PKA-dependent manner. These EVs exert paracrine tumor-suppressive effect by downregulating RAC2/PAK4 expression and inhibiting PI3K phosphorylation in neighboring pancreatic cancer cells, thereby attenuating malignant progression. Understanding this regulatory mechanism may contribute to improved risk stratification and surveillance strategy for IPMN-related pancreatic carcinogenesis.