Causal relationships between dietary habits and Barrett's esophagus risk: a univariable and multivariable Mendelian randomization study.

Aims: Dietary habits are reported to be associated with Barrett's esophagus (BE) risk; however, whether there is a causal relationship remains controversial. Here, we systematically examined the causal effects of genetically predicted dietary habits on BE risk through a Mendelian randomization (MR) analysis approach. Methods: Data for exposures were obtained from the UK Biobank (UKB), while the summary-level data for outcomes were obtained from a large sample-size GWAS meta-analysis. Genetic variants associated with 17 ordinary dietary habits at the genome-wide significance level were regarded as instrumental variables (IVs). Univariable and multivariable MR analyses were conducted to explore the causal relationships between dietary habits and BE risk. Sensitivity analyses were implemented to evaluate robustness of the results and determine the potential pleiotropy bias. Results: Univariable MR (UVMR) analysis showed that genetic predisposition to alcohol intake frequency, cooked vegetable intake, beef intake, bread intake, fresh fruit intake, salad/raw vegetable intake, and dried fruit intake were associated with BE risk, with all P values <0.05. After adjusting confounders, the effects of four dietary habits on BE risk persisted; multivariable MR (MVMR) analysis revealed that alcohol intake frequency (adjusted odds ratio (OR) = 1.74 (1.34, 2.27); P = 3.42 × 10-5) was causally associated with higher BE risk, the cooked vegetable intake (adjusted OR = 2.64 (1.16, 5.97); P = 0.02) had suggestively increased BE risk, while higher consumption of bread (adjusted OR = 0.54 (0.32-0.91); P = 0.02) and fresh fruit (adjusted OR = 0.34 (0.15, 0.77); P = 0.01) were suggestively associated with lower BE risk. Conclusions: These MR analyses demonstrate evidence of causal relationships between dietary habits and BE risk. These findings provide new insights into targeted dietary intervention strategies for BE prevention.

Analysis on the risk factors and establishment of a prediction model for primary non-response to the treatment of anti-tumor necrosis factor-α monoclonal antibody in Crohn′s disease patients / 中华消化杂志

Objective:To investigate the risk factors and establish a prediction model of primary non-response (PNR) to anti-tumor necrosis factor-α(TNF-α) monoclonal antibody in Crohn′s disease (CD) patients.Methods:From December 1, 2018 to July 31, 2022, 103 patients with CD treated with the anti-TNF-α monoclonal antibody in Renmin Hospital of Wuhan University were enrolled (modeling group), and at the same time, 109 patients with CD treated with anti-TNF-α monoclonal antibody in Zhongnan Hospital of Wuhan University were selected (validation group). The baseline clinical data of all the patients before the first treatment of anti-TNF-α monoclonal antibody were collected, which included C-reactive protein (CRP), the simplified Crohn′s disease activity index (CDAI), and modified multiplier simple endoscopic score for Crohn′s disease (MM-SES-CD), etc. Multivariate logistic regression was used to screen the independent risk factors of PNR in patients with CD treated with the anti-TNF-α monoclonal antibody, and to establish the nomograms prediction model. The area under the curve (AUC) of the receiver operating characteristic curve (ROC), the net reclassification index (NRI), integrated discrimination improvement index (IDI), and decision curve analysis (DCA) were used to evaluate the predictive efficacy and clinical application value of the prediction model. DeLong test was used for statistical analysis.Results:The results of multivariate logistic regression analysis showed that high level of CRP ( OR=1.030, 95% confidence interval (95% CI) 1.002 to 1.059), simplified CDAI ( OR=1.399, 95% CI 1.023 to 1.913), and MM-SES-CD ( OR=1.100, 95% CI 1.025 to 1.181) in baseline were independent risk factors of PNR in patients with CD treated with the anti-TNF-α monoclonal antibody ( P=0.033, 0.036 and 0.008). The results of ROC analysis showed that the AUCs of CRP, simplified CDAI, MM-SES-CD, and the prediction model in the modeling group and the validation group were 0.697(95% CI 0.573 to 0.821), 0.772(95% CI 0.666 to 0.879), 0.819(95% CI 0.725 to 0.912), 0.869 (95% CI 0.786 to 0.951) and 0.856 (95% CI 0.756 to 0.955), respectively. The AUC of the prediction model in the modeling group was greater than those of CRP and simplified CDAI, and the differences were statistically significant ( Z=3.00 and 2.75, P=0.003 and 0.006), while compared with MM-SES-CD and the validation group, the differences were not statistically significant (both P>0.05). However, compared with MM-SES-CD, the NRI and IDI of the prediction model in the modeling group were 0.205(95% CI 0.002 to 0.409, P=0.048) and 0.098(95% CI 0.022 to 0.174, P=0.011), respectively, suggesting that the predictive ability of the prediction model was better than that of MM-SES-CD. The results of DCA indicated that the prediction model had significant clinical benefits in both the modeling group and the validation group. Conclusions:A prediction model was successfully constructed based on the independent risk factors for PNR in patients with CD treated with the anti-TNF-α monoclonal antibody. After verification, the prediction model has good prediction performance and significant clinical benefits.