Using machine-learning algorithms to identify patients at high risk of upper gastrointestinal lesions for endoscopy

BACKGROUND AND AIM

Endoscopic screening for early detection of upper gastrointestinal (UGI) lesions is important. However, population-based endoscopic screening is difficult to implement in populous countries. By identifying high-risk individuals from the general population, the screening targets can be narrowed to individuals who are in most need of an endoscopy. This study was designed to develop an artificial intelligence (AI)-based model to predict patient risk of UGI lesions to identify high-risk individuals for endoscopy.

METHODS

A total of 620 patients (from 5300 participants) were equally allocated into 10 parts for 10-fold cross validation experiments. The machine-learning predictive models for UGI lesion risk were constructed using random forest, logistic regression, decision tree, and support vector machine (SVM) algorithms. A total of 48 variables covering lifestyles, social-economic status, clinical symptoms, serological results, and pathological data were used in the model construction.

RESULTS

The accuracies of the four models were between 79.3% and 93.4% in the training set and between 77.2% and 91.2% in the testing dataset (logistics regression: 77.2%; decision tree: 87.3%; random forest: 88.2%; SVM: 91.2%;). The AUCs of four models showed impressive predictive ability. Comparing the four models with the different algorithms, the SVM model featured the best sensitivity and specificity in all datasets tested.

CONCLUSIONS

Machine-learning algorithms can accurately and reliably predict the risk of UGI lesions based on readily available parameters. The predictive models have the potential to be used clinically for identifying patients with high risk of UGI lesions and stratifying patients for necessary endoscopic screening.

Prevalence of atrophic gastritis in southwest China and predictive strength of serum gastrin-17: A cross-sectional study (SIGES)

A hospital-based cross-sectional study in SIGES project was conducted during 2016.5-2017.5 in West China Hospital. It was aimed to observe the prevalence of atrophic gastritis (AG) in southwest China, and assess the diagnostic strength of serum gastrin-17 (G-17) in predicting AG in Chinese population. Asymptomatic healthy controls from health check-up, cancer-free patients with unspecific upper gastrointestinal symptoms, and histologically proven gastric cancer patients were eligible, if serum pepsinogen-I (PG-I), PG-II, and G-17 were detected. AG status was classified by the accredited cutoffs of PG-I (<70 ug/L) and PG-I/II ratio (<3). Totally, healthy controls (n = 9,425), symptomatic patients (n = 671) and gastric cancer patients (n = 305) were simultaneously observed, in which the prevalence of AG in southwest China were estimated as 15.9/1,000, 28.3/1,000, and 55.7/1,000 persons, respectively. The age-specific prevalence of AG in healthy controls showed a significantly uphill trend (p for trend <0.001). Higher level of serum G-17 was significantly associated with increased risk of AG in healthy population (15-30 pmol/L, aOR = 20.67, 95% CI 9.17-46.55; >30 pmol/L, aOR = 314.41, 95% CI 166.10-595.12). Throughout the progression of stomach diseases, the diagnostic strength of serum G-17 for AG showed a downhill trend across more advanced situations. In despite of that, serum G-17 displayed a good performance in predicting AG in the entire cross-sectional population (AUC = 0.92, 95% CI 0.89-0.94; SEN = 85.5%; SPE = 93.2%; LR+ = 12.55; LR- = 0.11). Population in southwest China had intermediate prevalence of AG, while the prevalence was increased over age or disease progression. High level of serum G-17 might be a reliable non-invasive measurement to predict AG in southwest Chinese population.

Serum Pepsinogen as a Predictor for Gastric Cancer Death: A 16-Year Community-based Cohort Study

GOALS

The purpose of this article is to validate the long-term association between initial serum pepsinogen (PG) measurements and subsequent gastric cancer-specific deaths from a long-term longitudinal cohort.

BACKGROUND

Endoscopic surveillance can be effective and efficient in reducing gastric cancer mortality if a biomarker such as serum PG is available to identify high-risk individuals and if the biomarker also is specific to gastric cancer risk.

STUDY

Between 1995 and 1998, a gastric cancer-screening program was conducted in a high-risk population: The first stage involved PG testing, and the second stage involved upper endoscopy. The outcome was gastric cancer death, which was monitored until December 31, 2010; results were expressed as the hazard ratio (HR) and corresponding 95% confidence interval (CI) using the Cox proportional hazards regression model. Other causes of death were used as comparators.

RESULTS

Among participants (n=3514) aged ≥30 years, 1682 (47.9%) were screened to determine serum PG levels. After 16 years of follow-up, 14 deaths from gastric cancer were documented. Multivariate analyses adjusted for age, sex, and Helicobacter pylori serological positivity showed that PG-I <30 μg/L and PG-I <30 μg/L or PG-I/II ratio <3 were significantly associated with the risk of gastric cancer death (HR, 3.27; 95% CI, 1.11-9.61 and HR, 3.45; 95% CI, 1.18-10.12, respectively). In contrast, there were no significant associations between PG and other causes of death, including neoplastic and non-neoplastic diseases.

CONCLUSION

This long-term cohort study shows the usefulness of PG measurement as a biomarker that is specific to the risk of gastric cancer death.

Inhibition of Vav3 gene can promote apoptosis of human gastric cancer cell line MGC803 by regulating ERK pathway

Previous studies proved that Vav3 gene was overexpressed in cancers. However, the molecular mechanism of Vav3 in apoptosis still keeps unclear; therefore, the relationship between Vav3 gene and apoptosis of gastric cancer (GC) was explored in the present study. Vav3-siRNA was transfected into MGC803 cells, and then cell activity and apoptosis rate were tested with MTT and FCM; apoptosis-related genes and proteins in MAPK signaling pathway were also tested. Results showed that Vav3 was overexpressed in GC than in adjacent normal tissues (all P < 0.05), and expression of Vav3 was related to degree of histological differentiation, cancer invasion depth, and lymphatic metastasis (Χ (2)  = 7.185, P = 0.007; Χ (2)  = 18.654, P < 0.001; Χ (2)  = 5.058, P = 0.025). Vav3 silencing inhibited activity of MGC803 cells, and apoptosis rate of cells was affected. Vav3-siRNA transfection led to changes of apoptosis-related genes such as Survivin, xIAP, Bcl-2, caspase-3, and Bax (all P < 0.01). After transfection, ratio of phosphorylation of ERK significantly reduced. We concluded that Vav3 inhibition can suppress cell activity and promote apoptosis by regulating the apoptosis-related genes through the ERK pathway.

Inhibition of gastric cancer cell growth and invasion through siRNA-mediated knockdown of guanine nucleotide exchange factor Vav3

Vav3, a Rho GTPase guanine nucleotide exchange factor, is associated with tumor growth, apoptosis, invasion and metastasis, and angiogenesis. However, the role of Vav3 in gastric cancer remains unclear. In this study, Vav3 expression was blocked by specific siRNA in gastric cancer cell line MGC803. MTT was used to assay cell proliferation activity; wound healing assay and transwell assay were applied to detect cell migration and invasion ability; and qRT-PCR and Western blot were employed to detect expression levels of Vav3 as well as proliferation, migration, and invasion-related genes. The results showed that Vav3 expression in gastric cancer tissues and cell lines was significantly upregulated and was higher than that in adjacent tissues of cancer and normal gastric mucosal cell lines. Vav3 knockdown inhibited proliferation, migration, and invasion of MGC803 gastric cancer cells. The expression of P21, P27, TIMP-1, and TIMP-2 was upregulated, while proliferating cell nuclear antigen, cyclin E1, matrix metalloproteinase (MMP)-2, and MMP-7 were downregulated by Vav3 knockdown in MGC803 gastric cells. In conclusion, Vav3 is involved in the proliferation, migration, and invasion of gastric cancer cell as a tumor oncogene.