Study population
We conducted a cross-sectional study of healthy adultswho participated in a health-screening program for disease prevention from September 2008 to December 2014 at the Health Promotion Center of the Samsung Medical Center, Seoul, South Korea. Data were collected from 1,808subjects who underwent Mini-Mental State Examination (MMSE), brain magnetic resonance imaging (MRI) including 3-dimentional (3D) volume images and esophagogastroduodenoscopy. The following subjects were excluded:11subjects under 45 years of age; 82 subjects with significant cognitive impairment which was definedaccording toMMSE scores below the 16th percentile of age- and education-matched normal population; 102 subjects whose education data were missing; 17 subjects with large brain lesions such as hemorrhage, ischemia, and mass;76subjects with missing data for CRP and 25 subjects with increased CRP (>1.0 mg/dL) which indicates superimposed active inflammation; 49 subjects with missing data for alcohol intake or smoking status. Thefinal sample size in this study was 1,446 subjects.
Data collection
The comprehensive health-screening program included demographic characteristics, anthropometric measurements, detailed physical examination, serum biochemical measurements, and a self-administered health questionnaire regardingyears of formal education, smoking status, alcohol consumption, medication use, and personal medical history such asdiabetes, hypertension, dyslipidemia, and cardiovascular disease. Smoking status was categorized into 3 groups including never, former, and current smoker. Alcohol consumption status was categorized into never drinker and drinker.Blood samples were collected from theantecubital vein after at least 10 hours of fasting.Detailed information regarding this screening program was previously provided[24].
Metabolic syndrome-related factorswere reviewed according tothe 2006 International Diabetes Federation (IDF) criteria for metabolic syndrome[25].We granted 1 point for each five factors of IDF criteria (1) body mass index (BMI)> 30kg/m2; (2) fasting plasma glucose >100mg/dL or on diabetes medication (3)blood pressure > 130/85 or on anti-hypertensive medication; (4)triglycerides>150mg/dL or on lipid lowering agent; (5)high-density lipoprotein (HDL) cholesterol< 40mg/dL for males, <50mg/dL for females or on treatment for dyslipidemia. Subjects were scored on a 0 to 5 scale for metabolic syndrome.
Assessment of H.pylori infection
The diagnosis of H. pylori infection was based on histological assessment. Board-certified gastroenterologists performed agastroendoscopyfor subjects who fasted overnight. Biopsy samples were taken from any region of the stomach and sent to the pathology department where the tissues were stained with hematoxylin and eosin and examined by qualified pathologists[26].
Measurement of brain cortical thickness
All subjects underwent a 3D volumetric brain MRI scan. An Achieva 3.0-Tesla MRI scanner (Philips, Best, the Netherlands) was used to obtain3DT1 turbo field echo MRI data.The following imaging parameters were included: sagittal slice thickness, 1.0-mm-thick sagittal slices with 50% overlap; no gap; repetition time of 9.9 milliseconds; echo time of 4.6 milliseconds; flip angle of 8°; and matrix size of 240 x 240 pixels reconstructed to 480 x 480 over a 240mm field of view.
The standard Montreal Neurological Institute image processing software (CIVET) was used to automatically processing ofT1-weighted MRIs to measure the cortical thickness. Native MRIs were first registered into a standardized stereotaxic space using an affine transformation[27]. Nonuniformity artifacts were corrected using the N3 algorithm, and the registered and corrected volumes were classified as white matter, gray matter, cerebrospinal fluid, and background using an artificial neural net classifier[28, 29]. The surfaces of inner and outer cortices were automatically extracted by deforming a spherical mesh onto the gray/white boundary in each hemisphere using the Constrained Laplacian-Based Automated Segmentation with Proximities algorithm[30, 31]. Cortical thickness was calculated as the Euclidean distance between the linked vertices of the inner and outer surfaces. To control for brain size, intracranial volume (ICV)was computed using classified tissue information and a skull mask, which was acquired from the T1-weighted image. Classified gray matter, white matter, cerebrospinal fluid, and background within the mask were transformed back into individual native space. To compare the thicknesses of corresponding regions among the subjects, the thicknesses were spatially registered on an unbiased iterative group template by matching the sulcal folding pattern using a surface-based registration that performs sphere-to sphere warping. We used SUMA[32]to parcellate lobar regions -frontal, temporal, parietal, and occipital lobes. Averaged values for the thickness of the whole vertex in each hemisphere were used for global analysis.
Statistical Analysis
To compare the difference in demographics of H. pylori-infected and-non-infected subjects, we used Student’st-test for continuous variables and chi-square test for categorical variables(Table 1).To evaluate the relationship between H.pylori infection and the braincortical thickness, we performed multiple linear regression analysis for each sex. Model 1 wasadjusted for age, ICV, years of education, alcohol intake, and smoking status (Table 2).Model 2 was further adjusted for CRP to Model 1. Model 3 was further adjusted for metabolic syndrome score to Model 1. Finally, Model 4 was adjusted for CRP and metabolic syndrome score to Model 1. For the analysis, H.pylori negative subjects were set as the reference group. Statistically significant cutoff value was defined as P-value<0.05. SPSS 25.0 (IBM, Armonk, NY, USA) was used for statistical analyses.
For evaluating the topography of cortical thickness differences associated with H.pylori infection, the MATLAB-based toolboxwas used[33].To blur each cortical thickness map, full-width half-maximum diffusion smoothing of 20 mm was used, resulting in increased signal-to-noise ratio and statistical power[34]. Linear mixed models were used, vertex by vertex, to analyze the localized differences and the statistical map of cortical thickness on the surface model.Each gender was analyzed after controlling for possible confounders as described in Models 1, 2, 3, and 4. The thresholds for statistical map results were determined using a false discovery rate (FDR) with a Q-value of 0.05 after pooling the P-values from regression analyses.