Background

Telomeres are repetitive nucleotides at the ends of eukaryotic chromosomes [1, 2] that are essential for genomic stability. Telomere length shortens with every cell division, due to the inability of DNA polymerase to extend DNA ends [1, 3]. Telomeres that shorten beyond a critical level cause cellular senescence [1, 3]; thus, telomere length has a considerable impact on human health [4].

Leukocyte telomere length (LTL) is commonly assessed in epidemiological studies [4]. LTL is affected by heredity [5, 6]; decreases with age [5, 7, 8]; is longer in females than males [7, 8]; correlates positively with paternal age at conception [6]; and varies by race, being longer, for example, in persons of African ancestry than individuals of European ancestry [4, 7, 9].

Shorter LTL was consistently linked to increased risk of atherosclerosis [10,11,12,13,14,15] and reduced survival [16,17,18]. The association between LTL and cancer risk is less consistent [19]. Some studies showed an increased risk for cancer in relation to short LTL [20, 21], while in other studies longer LTL was associated with increased cancer risk [22]. Given the importance of telomere length in human health, searching for modifiable determinants of LTL is highly desirable. Persons of low socioeconomic status displayed shorter LTL than did persons of high socioeconomic status [23]. Smokers were found to have shorter LTL than never or former smokers [24, 25]. Obesity [25] and higher body mass index (BMI) [26] were associated with shorter telomeres [26], while a positive association was found of physical activity with telomere length [27, 28]. An inverse association was shown between dietary caloric intake and LTL in men [29]. Persistent infections have been suggested to play a role in telomere length through the induction of oxidative stress and inflammation [30, 31]. For example, associations were reported of exposure to cytomegalovirus (CMV), herpes simplex virus type 1 (HSV-1), and human herpes virus 6 with greater LTL attrition in healthy adults [31], although these associations were not affected by systemic non-specific inflammatory markers such as C-reactive protein and interleukin 6 [31]. If infectious agents are truly associated with LTL, associations would be expected between tissue damage induced by these pathogens and LTL. Helicobacter pylori (H. pylori), a gram-negative bacterium that colonizes the stomach, can serve as a model to examine this hypothesis. H. pylori infection is acquired in childhood [32]. While H. pylori cause chronic gastritis in almost all infected persons, some present with peptic ulcers and gastric cancer in adulthood, especially those infected with strains that express cytotoxin-associated gene A (CagA) virulence antigen (reviewed in [33, 34]). Pepsinogen I (PGI) and PGII, pro-enzymes of pepsin, are secreted into the gastric lumen, and small amounts can be found in the serum [35, 36]. With increasing severity of H. pylori gastritis, serum PGI and PGII levels increase, but when atrophic gastritis ensues, the PGI level and the PGI:PGII ratio decrease. Serum PGs can predict atrophic gastritis and gastric cancer [37,38,39,40,41]. The objective of the current study was to examine associations of the sero-prevalence of H. pylori immunoglobulin G (IgG) antibody and serological evidence of atrophic gastritis with LTL in a general population sample.

Results

Description of the study sample

Overall, 934 participants (53.1% men) were tested for both H. pylori IgG antibody and LTL. The age at examination ranged from 27.0 to 78.0 years, with a mean of 52.0 years (standard deviation [SD] 13.9), which was similar among men and women. Most (64.2%) participants had not completed high school; this was more true for women (70.3%) than men (58.8%). Most participants (62.4%) defined themselves as traditional/secular: 67.5% and 56.6% in men and women, respectively. Overall, 82.0% of the participants reported being married; 94.2% and 68.1% among men and women, respectively. The overall prevalence of smoking was 26.0%: 40.8% in men and 9.0% in women; the respective figures for obesity were 44.1%, 31.9%, and 58.0% (Table 1).

Table 1 Characteristics of the study sample
Full size table

Mean LTL according to demographic and lifestyle factors

The LTL values ranged from 4.72 kb to 8.53 kb, with a mean of 6.76 kb (SD 0.61). The mean LTL was longer in women than men (p = 0.073). The mean LTL was longest in the youngest age group (27–34 years), 7.18 kb (SD 0.55), and decreased progressively in older age groups, reaching 6.40 kb (SD 0.55) at age 65–78 years (p < 0.001). All pairwise comparisons between the age groups were statistically significant by the Bonferroni test. The mean LTL differed significantly (p = 0.007) according to education, with a mean 6.72 kb (SD 0.58) among participants who had not completed high school, 6.83 kb (SD 0.61) in those who had completed high school (p = 0.055 by the Bonferroni test), and 6.87 kb (SD 0.71) in participants with academic education (p = 0.033 by the Bonferroni test). Persons who defined themselves as religious had shorter mean LTL than persons who defined themselves as traditional or secular (p < 0.001). Married individuals had longer mean LTL than unmarried ones: mean difference 0.11 (95% confidence intervals [CI] 0.01, 0.21), p = 0.039. Persons with obesity had shorter mean LTL, 6.71 kb (SD 0.60), than persons without obesity, 6.80 kb (SD 0.62), (p = 0.025). Physically active persons had longer mean LTL than non-active ones (p < 0.001); this association was found both for physical activity that was defined as sufficient and as high level. Mean LTL was not found to differ significantly according to the number of siblings (p = 0.2) and smoking (p = 0.6) (Table 2). Differences in the expected directions in mean LTL, according to age, religiosity, obesity, and high level of physical activity were observed in both men and women. In women, mean LTL differed significantly (p = 0.001) according to educational level, being the longest among those with academic education; but such difference was not significant (p = 0.3) in men (p for interaction 0.2). The difference in mean LTL according to marital status was significant in women only (p for interaction 0.009). The difference in mean LTL according to a modest level of physical activity level was significant among women only (p for interaction 0.07). No significant interactions were found between the other independent variables and sex (see Additional file 1).

Table 2 Mean leukocyte telomere length (kb) according to sociodemographic and lifestyle factors
Full size table

H. pylori sero-status, atrophic gastritis, and LTL

H. pylori IgG sero-positivity was found in 780/934 (83.4%) participants and atrophic gastritis in 81/927 (8.7%). In persons with a past H. pylori infection (lacked H. pylori IgG serum antibody), atrophic gastritis was found in 37 (4.0%), atrophic gastritis and H. pylori sero-positivity was evident in 44 (4.7%), and H. pylori sero-positivity without serological evidence of atrophic gastritis in 731 (78.9%), while 115 (12.4%) tested negative for H. pylori and lacked serological evidence for atrophic gastritis

The mean LTL was longer in H. pylori IgG sero-positive persons than in sero-negative ones: 6.78 kb (SD 0.59) vs. 6.65 kb (SD 0.66), p = 0.016; the difference was significant only when comparing H. pylori sero-positive persons who had CagA IgG serum antibody with those who were sero-negative (p = 0.018 by Bonferroni test). Participants who had serological evidence of atrophic gastritis had significantly shorter LTL (6.60 kb [SD 0.65]) than did participants without atrophic gastritis: 6.78 kb (SD 0.60) (p = 0.011). The shortest LTL was found in participants who had past H. pylori infection (sero-negative to H. pylori IgG antibody) and atrophic gastritis (6.40 kb [SD 0.67]); this compared to persons who were H. pylori sero-negative without atrophic gastritis (6.72 kb [SD 0.63]), those who were H. pylori sero-positive without atrophic gastritis (6.79 kb [SD 0.60]), and those who were H. pylori sero-positive with atrophic gastritis (p = 0.028, p = 0.001, and p = 0.037, respectively, by the Bonferroni test) (Table 3).

Table 3 Mean leukocyte telomere length (kb) according to H. pylori sero-status and serological evidence of atrophic gastritis
Full size table

A multiple linear regression model that included the variable H. pylori sero-status/atrophic gastritis and adjusted for age, sex, and religiosity was statistically significant (F statistic = 26.24, p < 0.001) with an adjusted R2 of 0.198 (Table 4). This model showed an inverse dose-response relationship between age and LTL, and significantly shorter LTL in men than in women; beta coefficient − 0.08 kb (95% CI − 0.15, − 0.01), p = 0.032. Compared to participants who were H. pylori sero-negative without atrophic gastritis, those with past H. pylori infection (sero-negative to H. pylori IgG antibody) and with atrophic gastritis had shorter LTL: beta coefficient − 0.21 kb (95% CI − 0.41, − 0.001), p = 0.049; the difference was not significant in H. pylori sero-positive persons either with (p = 0.10) or without atrophic gastritis (p = 0.3). The mean LTL was shorter in religious than traditional/secular participants; beta coefficient − 0.07 kb (95% CI − 0.14, 0.01), p = 0.078 (Table 4). The values of variance inflation factor (VIF) in this model ranged from 1 to 2, suggesting no collinearity.

Table 4 Multiple linear regression model of adjusted associations of demographic factors, H. pylori sero-status, and serological evidence of atrophic gastritis with leukocyte telomere length (kb)
Full size table

An additional model that included the variable H. pylori/atrophic gastritis sero-status and the following covariates; age, sex, religiosity, education, marital status, number of siblings, smoking, obesity, and physical activity showed similar results with regard to the associations of age, sex, and H. pylori/atrophic gastritis sero-status with LTL (see Additional file 2). However, no significant differences in LTL were observed according to obesity (p = 0.6), smoking (p = 0.7), physical activity (p = 0.5), marital status (p = 0.8), number of siblings, or education. The adjusted R2 of this model was 0.193, (F statistic = 13.86, p < 0.001). The VIF values ranged from 1 to 2, suggesting no collinearity. No significant interactions were found between H. pylori-atrophic gastritis sero-status with sex (p = 0.2), age (p = 0.12), education (p = 0.11), obesity (p = 0.2), physical activity (p = 0.12), and smoking (p = 0.8). Therefore, the interaction terms were excluded from the final model.

Discussion

We examined associations of H. pylori IgG antibody sero-positivity and serological evidence of atrophic gastritis with LTL in a general Arab population sample while assessing the role of sociodemographic and lifestyle factors.

As expected, a significantly shorter LTL was observed in older vs. younger participants and in men vs. women, thus confirming previous findings [5, 7, 8]. Exposure to H. pylori infection per se was not associated with shorter LTL. On the contrary, H. pylori IgG sero-positivity, especially CagA phenotype, was associated with longer LTL. Serological evidence of atrophic gastritis was associated with shorter LTL, the difference was driven by participants who tested negative to H. pylori but had atrophic gastritis. The longer LTL in H. pylori sero-positive participants can be explained by the higher prevalence of serological evidence of atrophic gastritis among H. pylori seronegative participants (24.1%) compared to those who were H. pylori seropositive, regardless of whether they were negative or positive for CagA IgG antibody: 5.7% and 6.0%, respectively, (p < 0.001) [42]. Likely, participants who were H. pylori sero-negative, but had atrophic gastritis, represent patients with the most severe form of gastric atrophy, which resulted in the loss of H. pylori infection [43]. Hence, H. pylori might negatively affect LTL only in a subset of infected persons, those with the most severe form of gastric atrophy. These findings confirm our general hypothesis that markers of tissue damage induced by H. pylori are related to shorter LTL. The longer duration of H. pylori infection observed in older persons is congruent with the acquisition of the infection in early childhood [32, 44], and its persistence [45], unless treated. H. pylori infection causes gastritis and typically, with aging, the severity of gastritis increases and atrophic lesions develop in the stomach. The prevalence of atrophic gastritis increases with age [42]. Hence, our findings might also suggest that infection of longer duration might decrease LTL; nonetheless, such interpretation should be made with caution, since information on the time in which the infection was acquired is not available.

On the first impression, our results might seem to contradict our expectations. In fact, our findings fit well with the natural history of H. pylori infection, indicating that despite the high prevalence of H. pylori infection of 44% globally [46], generally H. pylori does not cause disease, and only some infected persons develop peptic disease and gastric cancer (reviewed in [34]). The magnitude of the difference in LTL between persons with past H. pylori infection and atrophic gastritis and those who are H. pylori sero-negative and lacking atrophic gastritis was large (unadjusted mean difference − 0.32 kb [95% CI − 0.55, − 0.10]), p = 0.028. After adjustment for age, sex, and religiosity, the association was slightly attenuated but remained significant (beta coefficient − 0.21 kb [95% CI − 0.41, − 0.001]), p = 0.049. A case-control study of gastric cancer in Poland showed an increased risk for gastric cancer in relation to short LTL [47]. The same study showed among the control group, shorter LTL in persons positive than negative for H. pylori [47]. A study of non-neoplastic gastric mucosa from 106 cancer-free persons linked epigenetic changes, namely, H. pylori related hypermethylation of the promoter CpG island, with increased severity of gastritis and the development of atrophy (as measured by the PGI:PGII ratio); while shortened telomere increased the risk for hypermethylation [48]. Collectively, these and our observations shed light on the development of damage to the gastric mucosa in relation to H. pylori infection, in which telomere length shortening seems to play an important role. A small cross-sectional study of 163 US adults that examined associations of sero-positivity to four persistent pathogens (CMV, HSV-1, H. pylori, and Chlamydia pneumoniae) with total pathogen burden on LTL, showed reduced LTL in relation to CMV sero-positivity and increased pathogen burden in women (n = 100) but not in men (n = 63) [30]. LTL did not differ significantly according to H. pylori infection and no markers of atrophic gastritis were assessed [30]. Elsewhere, among ~ 400 participants aged 53–76 years, no significant association was found between CMV IgG sero-prevalence and LTL [49], but telomerase activity was reduced in relation to CMV positivity. Taken together, these and our findings suggest that exposure to persistent infections might play a role in LTL. Our findings suggest that strong specific inflammation of the stomach, as typically induced by H. pylori, is negatively associated with LTL. Longitudinal studies are needed to assess the directionality of the association and possible mechanisms between infections and LTL.

Reduced LTL in relation to obesity, and longer LTL in relation to physical activity were observed only in the bivariate analysis of the current study; such associations were not significant in the multivariable models. Smoking was not associated with LTL in our sample. These observations confirm our previous report [15] that was based on a sub-sample (n = 250) of the current cohort. Of interest, short LTL was strongly and positively associated with the prevalence of asymptomatic coronary atherosclerosis in that analysis [15].

Unlike previous findings [46], we found a negative association between religiosity and LTL, which was attenuated and became non-statistically significant in the multiple linear regression model that adjusted for age and sex. This suggests that the association between religiosity and LTL might result from confounding. Indeed, more women than men reported being religious (Table 1), and also older than younger participants (see Additional file 3).

Our study has some limitations. Using serum PGs to study atrophic gastritis might have limited sensitivity, which might result in non-differential misclassification of atrophic gastritis. The directionality of the associations of H. pylori sero-status and serological evidence of atrophic gastritis with LTL remain unknown because of the cross-sectional study design. Information on previous H. pylori eradication therapy was not collected, given the nature of our study, which utilized archived specimens and data that were obtained in a study on cardiovascular risk factors almost one decade ago. Persons with atrophic gastritis might change their dietary habits, yet such information was not available. Therefore, we cannot rule out the possibility of residual confounders.

The response rate to participate in the original study was 77% among Arabs; we cannot exclude the possibility that the non-responders might have different characteristics than the responders.

Our study has a number of strengths including the large general population sample with the representation of both sexes and various age groups, the attainment of findings of broad generalizability, the comprehensive assessment of demographic and lifestyle factors in addition to H. pylori and serum PGs, and the adjustment for confounders.

Conclusions

In this cross-sectional study, H. pylori IgG sero-positivity per se was not related to shorter LTL. However, persons lacking H. pylori IgG serum antibody with serological evidence of atrophic gastritis had shorter LTL than did those without atrophic gastritis, independent of other factors that might affect LTL.

Methods

Study design and population

We used archived anonymized specimens obtained in the framework of a cross-sectional study conducted during 2004-2008 among Jewish and Arab residents of Jerusalem. Details of the study design have been reported [15, 42, 50,51,52]. The current study was limited to the Arab participants for whom LTL measurement was performed.

The sampling frame included all permanent residents of East and West Jerusalem between ages 25–74 years, as recorded in the Israeli National Population Register. Random samples were drawn from the register for both population groups, stratified by sex and by 10-year age groups, 200 names in each stratum, for a total of 2000 names and addresses in each population. These were all invited to participate in the study. Individuals were ineligible if they were unable to provide informed consent, institutionalized, housebound, or had a severe illness; and women, if they were pregnant or gave birth within the 3 months preceding study initiation. The response rate among Arabs was 77% (n = 970) [15, 50]. We did not perform an a priori power calculation for the current study; we used all the available serum samples, 934/970 representing 96.3% of the Arab participants in the original study.

Data collection and definitions of the variables

Data were collected through personal interviews with the participants conducted during 2004–2008. Information was obtained on sex and age in years. Since the data were collected during a 4-year period, we considered age at examination, grouped here as 27–34, 35–44, 45–54, 55–64, and 65–78 years).

Self-reported education was classified into three categories: as having an academic degree, completed high school, and did not complete high school [42, 50, 52]. Marital status was defined as being married or not married (being single, divorced, or widowed). The variable religiosity was assessed in view of previous reports linking between religiosity and beneficial health outcomes [53,54,55], and reports on associations between religiosity and LTL [46]. Religiosity was defined based on the participants’ reply to a single question: “How do you define yourself?” The possible responses were very religious, religious, traditional, and secular. Since only two participants defined themselves as very religious and only 40 participants defined themselves as secular, the responses were grouped into two categories: religious vs. traditional/secular. Self-reported number of siblings was defined as having 0–3, 4–7, or ≥ 8 siblings. Smoking was classified as reported smoking of at least one cigarette daily vs. no smoking/other (i.e., smoking less than one cigarette/day). Physical activity was evaluated using the Multi-Ethnic Study of Atherosclerosis questionnaire [50], which assesses all-domains of physical activity including leisure, transport, work, and at home. Physical activity was defined following the World Health Organization (WHO) recommendation for physical activity in adults aged 18–64 years [56]. Participants were classified as sufficiently physically active if they met the WHO recommendation of doing at least 150 min of at least moderate intensity aerobic physical activity; 75 min of at least vigorous physical activity or an equivalent combination of moderate and vigorous intensity physical activity of at least 600 metabolic equivalents (METs) minutes weekly [56]. Participants were classified as highly active if they met the physical activity level associated with health benefits (i.e., doing at least 300 min of moderate-intensity aerobic activity throughout the week; or doing at least 150 min of vigorous-intensity aerobic activity; or an equivalent combination of moderate and vigorous intensity physical activity of at least 1500 METs minutes weekly). Height and weight were measured with light clothing and without shoes. Weight in kilograms was measured to the nearest 100 g using an analog scale. Standing height was measured to the nearest 0.1 cm. BMI was calculated as weight (in kg)/height (in meters [m])2. Obesity was defined as BMI ≥ 30 kg/m2.

Laboratory methods

Sera were tested for the presence of specific H. pylori IgG antibodies (Enzygnost® Anti-Helicobacter pylori II/IgG kit, Siemens Diagnostics Product GmbH, Marburg, Germany). The sensitivity and specificity of the kit are 94–98%. The presence of IgG antibody against recombinant CagA protein was measured in H. pylori-positive sera following an in-house ELISA protocol as previously described [42, 52, 57]. H. pylori sero-status was defined as (1) H. pylori negative if participants lacked H. pylori IgG antibody; (2) H. pylori-positive CagA negative, if they had H. pylori IgG antibody but lacked CagA IgG antibody; (3) H. pylori-positive CagA positive if they had both H. pylori and CagA IgG antibodies.

The level of serum PGI and PGII was quantified by ELISA (Biohit Inc., Helsinki, Finland) and the ratio of PGI:PGII was calculated. Serological evidence of atrophic gastritis was defined as a serum PGI level of < 30 μg/L or PGI:PGII ratio of < 3.0, as recommended by the manufacturer. Participants were also classified according to H. pylori IgG sero-positivity and serological evidence of atrophic gastritis as (1) H. pylori negative, no atrophic gastritis; (2) H. pylori positive, no atrophic gastritis; (3) H. pylori positive plus atrophic gastritis; or (4) H. pylori negative plus atrophic gastritis. Since atrophic gastritis is caused mainly by H. pylori infection and severe gastric atrophy results in loss of H. pylori infection [43], persons who tested negative to H. pylori but had atrophic gastritis were considered as having past H. pylori infection. All serological assays were performed in one laboratory at Tel University by an experienced technician who was masked to the LTL results and information on the other independent variables.

The measurement of LTL was performed using southern blot analysis of the terminal restriction fragment length at the laboratory of Professor Abraham Aviv, the Center of Human Development and Aging, Rutgers, The State University of New Jersey, New Jersey Medical School, Newark, USA [15, 49]. LTL (kb) was analyzed as a continuous variable expressed in kilobase pair.

Statistical analysis

The study sample (overall and by sex) was described using means (and standard deviations (SDs)) for continuous variables, and frequency and percentages for categorical variables. Differences in mean LTL (kb) were examined using the Student’s t test for dichotomous variables (i.e., sex, religiosity, marital status, physical activity, obesity, smoking) and one-way analysis of variance (ANOVA) for categorical variables with more than two categories (i.e., age, education, number of siblings, and H. pylori sero-status/atrophic gastritis). The Bonferroni post hoc test was used for pairwise comparisons to adjust for multiple comparisons. The variable LTL followed approximately a normal distribution. The assumptions of equal variance in Student’s t test and homogeneity of variance in ANOVA were examined and met. These analyses were performed in stratification by sex, given the documented difference between men and women in LTL [8]. Interactions between sex and the independent variables were assessed as an initial step before running a pooled multivariable model.

Multiple linear regression models were fitted to assess the associations between H. pylori sero-status/atrophic gastritis and LTL while adjusting for demographic and lifestyle factors. H. pylori sero-status/atrophic gastritis was enforced into the model as the main independent variable of interest. Other independent variables were selected to be included in the model based on prior knowledge; if the independent variables were associated with LTL (age, sex [5, 7, 8], religiosity [46], education as a measure of SES [23], smoking [24, 25], obesity [25], and physical activity [27, 28]), or with H. pylori infection (education and number of siblings [58]). Categorical variables were included in the model as dummy variables. Covariates were excluded from the analysis if they were associated with LTL in the model with p > 0.2, and if they yielded a change of less than 10% in the point estimate [59] of the association between H. pylori sero-status/atrophic gastritis and LTL (suggesting that they were not confounders), or in the adjusted R2. This resulted in two multivariable models, for each of them we reported the p value, F statistics, and adjusted R2. Interactions between H. pylori sero-status/atrophic gastritis and the other independent variables were assessed in the models. For each independent variable, we reported the beta (slope) coefficient (and 95% CI) that were obtained from these models. The assumptions of the linear regressions were assessed and met in all models. Collinearity between the independent variables was assessed using VIF. Statistical significance was set at p < 0.05. Data were analyzed using IBM SPSS version 25 (Armonk, NY, USA).