Active Helicobacter pylori Infection is Independently Associated with Nonalcoholic Steatohepatitis in Morbidly Obese Patients.

Nonalcoholic fatty liver disease (NAFLD) emerges as an important global burden and Helicobacter pylori infection (Hp-I) has been suggested as a risk factor of NAFLD, although controversy exists. This retrospective study aimed to investigate a potential impact of active Hp-I on NAFLD severity in morbidly obese patients, subjected to bariatric surgery and gastric biopsy for documentation of Hp-I. Of 64 eligible participants, 15 (23.4%) were diagnosed with active Hp-I, showing higher rates of nonalcoholic steatohepatitis (NASH) than those without Hp-I (86.7% vs. 26.5%, respectively; p < 0.001). Concerning histological lesions, steatosis grade (p = 0.027), ballooning (p < 0.001), lobular inflammation (p = 0.003), and fibrosis stage (p < 0.001) were also more severe in Hp-I positive patients. Likewise, liver function tests, insulin resistance, dyslipidemia, and arterial hypertension were significantly higher in Hp-I positive patients. Hp-I was independently positively associated with NASH (beta = 3.27; p = 0.002), severe NASH (beta = 2.37; p = 0.018), and the presence of fibrosis (beta = 3.86; p = 0.001) in a binary regression model, after adjustment for potential confounders. In conclusion, active Hp-Ι was independently associated with NASH and fibrosis, findings offering potential clinical implication.


Introduction
Nonalcoholic fatty liver disease (NAFLD), as the hepatic component of the metabolic syndrome (MetS), represents a leading cause of chronic liver disease; its estimated global prevalence is about 25% of the general population [1] whereas in morbidly obese individuals its prevalence might rise thyroid or adrenal function; kidney disease; pregnancy or lactation; use of glucocorticoids, insulin, orlistat, methotrexate, amiodarone, vitamin E, pioglitazone, intravenous glucose, or parenteral nutrition administration; any drug addiction; type 1 DM; pancreatitis; thrombotic disorders. Patients with T2DM were excluded if they were on thiazolidinediones or insulin treatment.

Data Collection and Extraction
Morbidly obese patients undergoing an elective bariatric operation were recruited. Specifically, the Hp-I status was evaluated by the diagnostic "gold standard" histology from gastric biopsies, obtained either preoperatively (in terms of a routine esophagogastroduodenoscopy) or intraoperatively. Histological diagnosis of Hp gastritis was made on 5 µm formalin-fixed, paraffin-embedded tissue sections by means of Hematoxylin and Eosin staining. A modified Giemsa stain was also used to highlight the Hp bacteria.
Liver specimens were obtained intraoperatively during laparoscopic bariatric surgery, from liver segment II or, less commonly, segment III. Grading and staging of NAFLD was based on the NASH Clinical Research Network scoring system [32]; for the discrimination between NAFL and NASH, both NAFLD activity score (NAS) [32] and fatty liver inhibition of progression (FLIP) were used, the latter introduced in morbidly obese populations [29]. Severe NASH was defined as steatosis, activity, and fibrosis (SAF) scoring system ≥3 and/or F ≥ 3, as elsewhere recommended [33].
Patients' records at the time of admission to the ED were stored in the clinical application E.care for Windows (E.care BVBA, ED 2.1.3.0, Turnhout, Belgium). E.care offers the advantage of instantaneous recall of medical reports, and other relevant data, while multiple filters of E.care application can be applied. Patients' records from ED were extracted to an Excel sheet (Microsoft ® Excel for Mac 2019, Microsoft Corporation, Redmond, WA, USA) with the use of appropriate filters. Eligibility of the retrieved patients was evaluated by two investigators (M.D. and S.S.), following the inclusion and exclusion criteria. Selected data were validated by D.S.S. In cases of conflict, a consensus was met by the intervention of a senior author (A.E.). The following parameters were extracted: (a) Demographics and anthropometric (age, gender, body-mass index (BMI)); (b) histological findings of gastric and liver biopsies: Hp-I positivity, steatosis grade, lobular inflammation, ballooning degeneration, and fibrosis; (c) MetS components, i.e., IR, dyslipidemia, hypertension, and T2DM, as recommended by the Expert Committee on the Diagnosis and Classification of Diabetes [34]; in this regard, dyslipidemia was defined as triglyceride levels ≥150 mg/dL (1.7 mmol/L) or LDL-C levels ≥ 100-160 mg/dl (2.58-4.13 mmol/L, depending on other risk factors) or HDL-C levels < 40 and 50 mg/dl (1.03 and 1.29 mmol/L) in men and women, respectively, or treatment with hypolipidemic medication(s); arterial hypertension was defined as systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 85 mmHg or treatment with antihypertensive medication(s); (d) laboratory tests: Fasting glucose and fasting insulin (for the calculation of IR, using the homeostasis model assessment IR (HOMA-IR) [35]), glycated hemoglobin A1c (Hba1c), liver function tests (aspartate aminotransferase (AST), alanine aminotransferase (ALT), gamma-glutamyl transferase (GGT), bilirubin).

Statistical Analysis
Data are presented as mean ± standard deviation (SD) and percentages, for continuous and categorical variables, respectively. The normality of distributions of continuous variables was tested with the Kolmogorov-Smirnov test. The comparisons of continuous and categorical variables were performed with the Mann-Whitney test and chi-square test (or the Fischer exact test), respectively. Binary logistic regression analysis was performed to investigate whether Hp-I was independently associated with NASH, severe NASH and fibrosis, because they are considered the main histological endpoints. As independent variables, we entered parameters related to MetS (BMI, hypertension, dyslipidemia, HOMA-IR), as well as age and gender. Statistical analysis was performed with SPSS 21.0 for Macintosh (IBM Corp., Armonk, NY, USA). Significance was set at p < 0.05 (two tailed).
Comparative data between Hp-I positive [Hp(+)] and negative [Hp(−)] patients are presented in Table 1. Hp(+) and Hp(−) groups were of similar gender age, and BMI. AST, ALT, GGT, bilirubin, triglycerides, and HOMA-IR were higher in Hp(+) than in Hp(−) group, whereas total cholesterol, HDL-C, and LDL-C were not statistically different between groups. Higher rates of hypertension (p = 0.041) and triglyceridemia (p = 0.039), but not prediabetes/diabetes, were observed in Hp(+) than Hp(−) group. Regarding both FLIP and NASH classifications, higher rates of NASH were observed in Hp(+) than in Hp(−) group (Table 1). Likewise, higher rates of severe NASH were observed in Hp(+) than in Hp(−) group. Concerning separate histological lesions, histological severity was more prominent in Hp(+) than Hp(−) patients in steatosis grade, ballooning, lobular inflammation, and fibrosis (Table 1, Figure 1). Representative histological images from the included patients are illustrated in Figure 2. When the analysis was repeated in women (n = 47), the results remained essentially unchanged. However, the analysis in men was avoided, mainly owing to the small sample of men (n = 17), which might have led to insecure conclusions. J. Clin. Med. 2020, 9, x FOR PEER REVIEW 4 of 14
Comparative data between Hp-I positive [Hp(+)] and negative [Hp(−)] patients are presented in Table 1. Hp(+) and Hp(-) groups were of similar gender age, and BMI. AST, ALT, GGT, bilirubin, triglycerides, and HOMA-IR were higher in Hp(+) than in Hp(−) group, whereas total cholesterol, HDL-C, and LDL-C were not statistically different between groups. Higher rates of hypertension (p = 0.041) and triglyceridemia (p = 0.039), but not prediabetes/diabetes, were observed in Hp(+) than Hp(−) group. Regarding both FLIP and NASH classifications, higher rates of NASH were observed in Hp(+) than in Hp(−) group (Table 1). Likewise, higher rates of severe NASH were observed in Hp(+) than in Hp(−) group. Concerning separate histological lesions, histological severity was more prominent in Hp(+) than Hp(−) patients in steatosis grade, ballooning, lobular inflammation, and fibrosis (Table 1, Figure 1). Representative histological images from the included patients are illustrated in Figure 2. When the analysis was repeated in women (n = 47), the results remained essentially unchanged. However, the analysis in men was avoided, mainly owing to the small sample of men (n = 17), which might have led to insecure conclusions.   Binary regression analysis was performed with FLIP as dependent variable in a subset of patients (NALF vs. NASH; n = 55), after the exclusion of those without NAFLD (n = 9). The results of this analysis are presented in Table 2. NASH remained independently positively associated with Hp-I (p = 0.002) after adjusting for potential confounding. Next, regression analysis was performed with severe NASH as dependent variable (No vs. Yes) in the sum of patients. Again, severe NASH remained independently positively associated with Hp-I (p = 0.018; Table 3). In this model, age and BMI were also independently positively associated with severe NASH. Likewise, when fibrosis (F0 vs. F1-3) in the sum of patients was selected as dependent variable, the presence of fibrosis remained independently positively associated with Hp-I (p = 0.001; Table 4). In this model gender was also independently associated with the presence of fibrosis (i.e., men had independently higher risk of fibrosis).

Discussion
This study favors an association between active Hp-I and histological severity of NAFLD in morbidly obese patients subjected to bariatric surgery. Specifically, higher rates of NASH, as well as hepatic steatosis, inflammation, and fibrosis were observed in Hp(+) than Hp(−) patients, findings warranting further investigation. Notably, Hp-I was independently positively associated with NASH, severe NASH, and hepatic fibrosis, the latter considered as the main histological prognostic factor [36].
Importantly, the histological diagnostic "gold standard" for both main variables of interest (active Hp-I and NAFLD) were used, which is scarce in the literature. More specifically, this is the second study, after the study of Lecube et al. [14] investigating the association between Hp-I and NAFLD by using both gastric and liver biopsies, in which liver biopsy was performed in a subset of patients (22% of patients subjected to gastric biopsy). In that study, higher steatosis rates were observed in Hp(+) patients, as in our study, but, unexpectedly rates of NASH were higher in Hp(−) patients. Moreover, contrary to our finding, the authors reported similar overall NAFLD rates between Hp(+) and Hp(−) patients [14]. There is no secure conclusion about the controversy of our with Lecube et al. [14] study; population and/or methodological differences may exist, thus warranting more studies, specifically designed to this aim.
The association between Hp-I and components related to IR or MetS were also investigated in this study as secondary aims. Higher liver function tests, triglycerides, HOMA-IR, and higher rates of arterial hypertension were observed in Hp(+) compared with Hp(−) patients. In a systematic review we performed almost 10 years ago, we have proposed a positive association between Hp-I and IR, in terms of HOMA-IR [44]. Although controversy still exists, similar results to this study have provided a meta-analysis between Hp-I, MetS, and MetS-related comorbidities [45].
The observational nature of this study cannot show causality or direction of the relationship, i.e., to answer the question whether Hp-I is another pathogenetic factor of NAFLD or, inversely, NAFLD makes affected individuals more susceptible to Hp-I. However, pathogenetic mechanisms between Hp-I and NAFLD may be hypothesized. Hp-I related gastrointestinal epithelium disruption, microbiota, and their metabolites translocation into portal circulation, activation of inflammation via toll-like receptors signaling in hepatocytes, may be some contributors linking Hp-I with the development and progression of NAFLD [3,46]. In this regard, human β-defensin-1, possibly serving as a biomarker of bacterial translocation in chronic liver disease, is induced by Hp-I. Hp-I also contributes to a low-grade inflammation by inducing the release of vasoactive and proinflammatory molecules, including interleukin (IL)-1, IL-6, IL-8, IL-10, IL-12, tumor necrosis factor-α, interferon-γ, eicosanoids (leukotrienes, prostaglandins), and acute phase proteins (C-reactive protein and fibrinogen), which are implicated in the pathophysiology of IR syndrome and NAFLD. Moreover, Hp-related induction of oxidative stress, atrophic gastritis-associated with vitamin B12/folate deficiency, apoptosis, and the possible downregulation of adiponectin, constitute other pathogenetic links between Hp-I and NAFLD [3,46]. Furthermore, Hp-related galectin-3 connected with MetS may play a role in the pathophysiology of MetS-related NAFLD and its extrahepatic complications, including chronic kidney, cardiovascular, and brain disorders; the number of galectin-3 positive liver cells is associated with NAFLD severity; galectin-3 binding protein is a potential marker of disease; and galectin-3 inhibitors may exhibit potential efficacy in advanced fibrosis/cirrhosis due to NASH. Finally, Hp-related MetS conditions are associated with dysmotility-induced microbiota dysbiosis, particularly small intestinal bacterial overgrowth (SIBO) playing a role in the pathogenesis of NAFLD; SIBO exhibits an effect on the structural and functional issues of the liver leading to greater prevalence of NAFLD. Notably, SIBO prevalence is higher in morbidly obese patients than in healthy populations and is connected with severe hepatic steatosis [47]. However, further large-scale relative studies are needed to elucidate in depth the aforementioned proposed pathogenetic mechanisms.
Although the use of both liver and gastric biopsies is a strength of this study, there are also certain limitations. First, the retrospective nature of the study renders it susceptible to record and recall bias. Second, as mentioned above, a cause-effect relationship cannot be shown. Third, the sample size was small, although it was sufficient to show robust statistical significance in major endpoints. Furthermore, information on previous use of proton pump inhibitors before obtaining gastric biopsies was not available. Last, the duration of Hp-I, which might have affected the results, remains unknown.

Conclusions
This study favors an association of active Hp-I on NAFLD severity in morbidly obese patients subjected to bariatric surgery. Most importantly, active Hp-I was independently associated with NASH and the presence of fibrosis. These findings warrant specifically designed clinical studies on the association between active Hp-I and NAFLD severity in obese, morbidly obese, and lean populations, as well as mechanistic studies investigating the potential pathogenetic links.