Assessment of the diagnostic accuracy and relevance of a novel ELISA system developed for seroepidemiologic surveys of Helicobacter pylori infection in African settings

Abstract

Beside diagnostic uncertainties due to the lack of a perfect gold standard test for Helicobacter pylori infection, the diagnosis and the prevalence estimation for this infection encounter particular challenges in Africa including limited diagnostic tools and specific genetic background. We developed and evaluated the accuracy of an enzyme-linked immunosorbent assay (ELISA) system tailored for H. pylori genetics in Africa (HpAfr-ELISA). Strains belonging to main genetic populations infecting Africans were exploited as sources for whole-cell antigens to establish in-house the ELISA system. A phase II unmatched case-control study explored the diagnostic accuracy of the HpAfr-ELISA using a training set of samples collected from dyspeptic patients from Kinshasa, the Democratic Republic of Congo (DRC) who had been tested with invasive standard tests (i.e., histology, culture, and rapid urease test) in 2017. Then the assay was cross-validated through a community-based survey assessing the prevalence of H. pylori and associated factors in 425 adults from Mbujimayi, DRC in 2018. Bayesian inferences were used to deal with statistical uncertainties of estimates (true prevalence, sensitivity, and specificity) in the study population. At its optimal cut-off-value 20.2 U/mL, the assay achieved an estimated sensitivity of 97.6% (95% credible interval [95%CrI]: 89.2; 99.9%) and specificity of 90.5% (95%CrI: 78.6; 98.5). Consistent outcomes obtained at repeated tests attested the robustness of the assay (negative and positive agreements always > 70%). The true prevalence of H. pylori was estimated 53.8% [95%CrI: 42.8; 62.7%]. Increasing age (adjusted odds ratio [aOR] > 1.0 [95% confidence interval (CI): > 1.0; 1.1]; p<0.001), overcrowding households (aOR = 3.2 [95%CI: 2.0; 5.1]; p<0.001), and non-optimal hand hygiene (aOR = 4.5 [95%CI: 2.0; 11.4]; p = 0.001) were independently associated with the H. pylori-seropositivity. The novel ELISA system has demonstrated good diagnostic accuracy and potential usefulness for management and mitigation strategies for H. pylori infection in African settings.

Introduction

Helicobacter pylori is a Gram-negative, spiral shaped, flagellated microaerophilic bacterium that infects half of living humans worldwide [1]. This infection is the major cause of a wide spectrum of diseases including non-immune chronic gastritis, peptic ulcer diseases, and gastric cancer [1,2]. Several risk factors play an important role in H. pylori infection including gender, age, and low socioeconomic status [3]. However, a substantial epidemiological variation and an exceptionally wide genetic diversification of this infection exist in different regions and human populations [1,3,4]. Strategies for effectively addressing this public health issue including the individual diagnosis and mass screening should thus be adjusted to population background.

H. pylori infection is diagnosed by several methods including invasive (e.g., culture from gastric biopsies, histology, and rapid urease assay) and noninvasive tests (e.g., PCR on stool samples, urease breath test, rapid urine test, serological test, and stool antigen test) [5–7]. Each of these tests carries proper imperfections and advantages but none of them fulfills the statistical requirements for perfect a gold standard test which is a reference and error-free diagnostic test [6,7]. Consequently, all tests have a potential for misclassifications of infected and non-infected people that can lead to inappropriate clinical and public health decisions. Unlike invasive tests, noninvasive ones are widely used for epidemiological investigations [7]. Serological tests such as Enzyme-linked immunosorbent assays (ELISA) are the only diagnostic methods that are able to detect past exposure to H. pylori [8]. Basically, the ELISA technique is relatively simple, reliable, fast, inexpensive, and has interesting high-throughput capabilities [5,6,8]. Several ELISA kits for H. pylori infection are commercially available [9]. For all these reasons, this technique is highly attractive for screening of large communities to evaluate prevalence including within regions with limited resources. However, the ELISA for H. pylori needs adjustment for specific regions and populations as infections in varied areas have clearly indicated different immunogenic properties [9–13]. Moreover, the nature of antigens from the same isolates plays a significant role in the improvement of ELISA sensitivity and specificity [14]. H. pylori species displays a diverse protein profile (e.g., CagA, VacA, UreaB, OipA, Heat shock protein B, Flagellar proteins) with more or less enhanced antigenic properties which are putative components of vaccines or ELISA systems [15–18]. During the colonization of the human stomach by H. pylori, these antigens activate humoral and cell mediated immune response with elevated titers of serum immunoglobulin predominantly of IgG class [15,17]. Whole-cell antigens covering a broad range of bacterial antigens have been found to be more immunogenic than isolated antigenic components [17].

Otherwise, compared to other continents, Africa has recorded the highest prevalence of H. pylori infection; however the epidemiology of this infection remains relatively less explored at community levels [3,4,19]. Africa is colonized by H. pylori bacterial strains that are the most genetically distant and structurally divergent from remaining species populations [20,21]. Beyond diagnostic uncertainties due to the lack of a perfect gold standard test, the clinical diagnosis and the prevalence estimation of H. pylori are surrounded with significant operational challenges in Africa. These include restricted access to diagnostic tools such as gastrointestinal endoscopy and H. pylori culture facilities, limited number of trained operators, specific genetic background, and huge favoring conditions for infection transmission [19]. Regarding ELISA systems, it is likely that available commercial kits that have essentially been optimized and evaluated in developed countries may not be suitable for Africa. We thus developed in-house a novel ELISA system (i.e., HpAfr-ELISA) based on H. pylori whole-cell antigens retrieved from bacterial isolates representing the main genetic populations infecting humans in Africa. The goal was to provide a serodiagnostic and seroepidemiological tool for H. pylori infection in Africa. This study aimed to determine the diagnostic accuracy and robustness of this tool in compliance with the evidence-based architecture of diagnostic research [22]. The diagnostic relevance of this system was further explored during a real-world community-based survey conducted for estimating the prevalence of H. pylori infection and for identifying its most predictive factors in the Democratic Republic of Congo (DRC), Middle Africa.

Methods

This study is reported following the “Strengthening the Reporting of Observational studies in Epidemiology” (STROBE) statement guidelines [23] and the STROBE checklist is provided in supporting information (S1 Table).

Ethics statement

All study subjects provided their informed consent for inclusion before participation in the study. A formal consent was thus obtained written. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethical Review Boards of the University of Mbujimayi in DRC (N#33/MREC/UM/ETK/GDT/2018).

Flowchart for the study methodology

The flowchart in Fig 1 displays the main methodological steps of this study.

[Figure omitted. See PDF.]

Fig 1. Flowchart for the study methodology.

This study followed three main steps including the development of the new ELISA assay followed by its analytical validation using a training set of clinical samples (Cases and Controls), and, finally, its implementation during a real-world epidemiological survey.

https://doi.org/10.1371/journal.pntd.0009763.g001

Study area, population and design

This is a phase II evaluation of a newly developed diagnostic test. First, the diagnostic accuracy of the test was evaluated experimentally through an unmatched case-control analysis performed using a training set of equal numbers of H. pylori-positive cases (n = 30) and H. pylori-negative controls (n = 30) obtained during a hospital-based survey on dyspeptic patients who underwent upper gastrointestinal examination by endoscopy in 2017 in Kinshasa, the capital city of The Democratic Republic of Congo (DRC) [24]. This sample size was expected to be a priori sufficient for reliably detecting an Area Under the Curve (AUC) of ≥ 0.70 as statistically different from 0.50 with an 80% power (α = 0.05; 2-sided test) [25]. Then, the test accuracy was cross-validated using samples (cross-validation set) from a mass community-based survey carried out in August 2018 in the Health Area of Kizito (coordinates: latitude, 6°6′S to 6°7′S; and longitudes, 23°37′E to 25°38′E) that is located on the outskirts of Mbujimayi, in the center of the DRC. This cross-sectional survey recruited participants at household level. Targeted households were randomly sampled using a frame elaborated based on satellite images captured over the study area with Google Earth (www.google.com/intl/eng/earth). Inclusion criteria were that the respondent was an adult (≥18 years old) residing in the study area and who provided an informed consent. Only one or two potentially eligible participants could be recruited from each household on a first-come, first-included basis. Exclusion criteria comprised those from whom a good quality biological sample or a full interview could not be obtained among potentially eligible subjects. A minimum sample size of 384 participants was deemed sufficient for this survey (based on a worldwide expected prevalence of 50% [4] with a precision of 5% and a type I error of 5%).

Data collection and definition of variables

A piloted standard questionnaire was used to collect socio-demographic information and dyspepsia symptoms in each study participants. Dried blood spots (DBS) were sampled using Whatman 903 filter card (GE Healthcare Ltd., UK) for biological testing. The variable of interest was the level of anti-H. pylori IgG. Participants were considered as being H. pylori-seronegative or -seropositive respectively when the anti-H. pylori IgG level was ≤ or > the best cut-off-value of the assay. Types of housing (i.e., structure of walls, floor, and roof) were used to generate a wealth and ownership categorization by Principal Component Analyses (PCA) rating participants into 2 socioeconomic levels: “more economically disadvantaged” and “less economically disadvantaged”. An index reflecting the crowding of residences was defined based on the number of household members sharing the same sleeping room. “Overcrowded households” were thus considered as those with an index of occupancy higher than the median index in the study population. The level of hand hygiene was appreciated as optimal or non-optimal based on the reported frequency of hand washing after being to toilets (i.e., corresponding to “always” or to one of the following “occasionally”, “sometimes”, “frequently”, and “usually”). Dyspepsia symptoms and their severity were evaluated using three clinical scores as established in previous studies: i) the Dyspepsia Symptom Severity Index (DSSI) with its three categories of symptoms referred to as “Dysmotility-like”, “Reflux-like”, and “Ulcer-like” [26]; ii) the Glasgow Dyspepsia Severity Score (GDSS) [27]; and iii) the Global Overall Symptom (GOS) Scale that assess the impact of symptoms on people’s quality of life [28]. The dyspepsia syndrome was defined based on the Rome IV criteria as any combination of 4 symptoms: postprandial fullness, early satiety, epigastric pain, and epigastric burning that are severe enough to interfere with the usual activities and occur at least 3 days per week over the last 3 months with an onset of at least 6 months in advance [29].

DBS samples processing

DBS samples were processed following the protocol developed by Grüner et al. [30]. Briefly, through a skin puncture made on a finger, five drops of blood were gently transferred to each of the circles of a Whatman 903 filter card. The filter card was put upon collection on a clean paper towel in a biohazard safety box to be dried at ambient temperature for up to 4 hours. Then, a single gas-impermeable zipper bag with a desiccant was used for storage of each filter card at ambient temperature. All DBS samples were transported to Oita University in Japan for laboratory analyses. Before testing, DBS were punched out from cards, transferred to a 24-well plate and eluted overnight on a shaker using 500 μL solution of phosphate-buffered saline (PBS, pH 7.2–4, Wako, Japan) containing 0.05% Tween 20 and 0.08% sodium azide. Eluted DBS were used for HpAfr-ELISA testing of H. pylori.

Development of the HpAfr-ELISA and laboratory testing

Anti-H. pylori IgG antibodies were measured using the HpAfr-ELISA assay, an indirect ELISA system developed in-house. This ELISA system immobilizes antigens from whole cells of four representative H. pylori strains isolated from Congolese patients in 2017 [24]. These isolates were selected based on their genetic background as they clustered in hpAfrica1 (two strains) and hpNEAfrica (two strains) (S1 Fig)–the two main genetic populations of H. pylori species colonizing the Sub-Saharan Africa [21,31,32]. To develop this assay, pure bacterial cells of each strain were suspended well in PBS and treated with 1% octyl-glucoside and 1mM Phenylmethylsulfonyl fluoride (PMSF) protease inhibitor (ThermoFisher Scientific, USA) in PBS. Then, 2,990 μL of each whole cell antigens lysate solution was dialyzed for 12 hours at 4°C using A-Lyzer 10K Dialysis Cassettes G2 (ThermoFisher Scientific, USA) in 2L of PBS. Using the BCA microassay (TaKaRa bio, Japan) with bovine serum albumin (BSA) as standard, the antigens concentration was determined in 10 μL of each lysate solution before and after dialysis. An equimolar mixture of antigen solutions extracted from each H. pylori isolate was prepared and used to coat a 96-wells Maxisorp ELISA plate (Immobilizer-Amino Plate, UK). Each well was thus coated with 100 μg of antigens mixture in 100 mM Na2CO3-NaHCO3 buffer (pH 9.6) overnight and washed three times with PBS containing 0.1% Tween 20 (PBS-0.1T). After blocking ELISA plates with 0.2% BSA in PBS, 1:1,000 eluted DBS samples were reacted for 30 minutes. A solution of 1:80,000 peroxidase-conjugated secondary anti-human IgG antibodies (Jackson Immuno Research Labs, USA) was added for 30 more minutes to each reacting well after washing the plates with PBS-0.1T. The final detection was performed with POD/TMB substrate (Nacalai Tesque, Inc., Japan). This ELISA system was optimized and calibrated using the training set of blood samples from Congolese patients whose H. pylori status was already established using culture, histology or rapid urease test applied as previously described (S2 Table) [31,33,34]. Polyclonal rabbit anti-Helicobacter pylori IgG (Dako, Danmark) with detection by 1:80,000 peroxidase-conjugated secondary anti-rabbit IgG antibodies (Jackson Immuno Research Laboratories, Inc; USA) was used as an internal positive control while water, PBS and carbonate buffer were set as negative controls during all reactions. We defined that 1 U/mL anti-H. pylori antibodies from human was comparable to 1ng/mL anti-H. pylori rabbit IgG.

Data analysis

The data were entered into Microsoft Excel spreadsheets (Microsoft Corp., Redmond, WA, USA, 2010) and analyzed using R software version 4.0.2 [35]. Data manipulations were performed with the tidyverse package [36]. Principal Component Analyses (PCA) were conducted using FactoMineR and Factoextra packages [37,38]. The AUC and ROC (Receiver Operating Characteristics) curves were estimated, analyzed, compared, and applied to determine the best cut-off-value of the assay using pROC package [39]. The DeLong’s test was used to evaluate the difference in AUCs of two correlated ROC curves. Qualitative variables were compared between H. pylori-seropositive and–seronegative groups using the Chi-square test or the Fisher’s exact test when appropriate. In contrast, quantitative variables were compared using parametric the Wilcoxon rank sum test with continuity correction after testing non-parametric conditions with the Shapiro-Wilk test and the Levene test. The reliability of the ELISA was assessed on both the training and the cross-validation sets through the Cohen’s Kappa and the Kendall’s W statistics calculated with KappaGUI and synchrony packages [40,41]. The 95% confidence interval (95%-CI) of proportions was estimated based on Clopper-Pearson exact interval. Uni- and multivariable logistic regression models were fitted to assess the association between different factors and the H. pylori-seropositivity. The final multivariable model was selected based on Akaike’s Information Criterion (AIC) values estimated for different intermediates models using both a backward and forward approach. For all the statistical tests applied, the value of p<0.05 was considered as the statistical significance threshold. Bayesian methods were applied for inferring latent estimates (e.g., true prevalence, sensitivity, and specificity) in the study population as updated posterior probabilities sampled through predefined probability distributions and Markov Chain Monte Carlo (MCMC) simulations [42–48]. The Bayesian model was fitted using prevalence package [44]. The median of each parameter was used to describe the uncertainty with a 95% credible interval (95%-CrI) based on the Highest Density Interval (HDI). At four parallel MCMC chains, each for 10,000 iterations for all the models with a burn-in period of 10,000 iterations, were performed. The parameters of each individual model were adjusted until the convergence of MCMC runs could be attested by Gelman and Rubin’s diagnostics measure (close to 1) [49]. The details of the Bayesian model fitted are presented in S1 File.

Results

Performance and robustness of HpAfr-ELISA

Average levels of anti-H. pylori IgG antibodies obtained from repeated testing (three times) on the training set with the HpAfr-ELISA were determined for individual participants. Hence, samples from H. pylori positive patients (i.e., based on standard invasive tests) had significantly higher anti-H. pylori IgG levels than those from negative ones (45.0 interquartile range [IQR]: 75.2) and 11.9 (IQR: 7.3) U/mL; p = <0.001) (Fig 2). The ROC curve displayed an AUC of 97.6% climaxing at its best cut-of-value 20.2 U/mL of anti-H. pylori IgG with corresponding sensitivity 96.7% and specificity 90.0% (Fig 3; Fig 4). To assess the robustness of the assay, we considered individual outcomes at repeated testing of both the training set and a larger dataset from a community-based survey (cross-validation set). Repeated training tests thus indicated no significance difference in the ROC curve (p>0.05) (Fig 5). Furthermore, based on the best cut-off-value of the ROC/AUC, the ELISA indicated high indices for qualitative agreement of the outcomes during repeated tests on both the training samples (Kappa = 0.833; PA = 0.915; NA = 0.918) and the cross-validation sets (Kappa = 0.789; PA = 0.881; NA = 0.908) (Table 1). Likewise, significantly high indices for quantitative agreement were noted with both sets (W = 0.956 to 0.981; p = 0.001 and W = 0.631; p = 0.001) (Table 2). As part of cross-validation, we assessed the performance of the assay on the dataset from the community-based survey but in the absence of a perfect gold standard test. Therefore, stochastic simulations under Bayesian assumptions were performed with prior sensitivity and specificity defined as Beta distribution of results obtained on the training set (S1 File). The latent sensitivity and specificity of HpAfr-ELISA were thus respectively estimated 97.6% with a 95% credible interval (95%CrI) from 89.2 to 99.9%, and 90.5% with 95%CrI from 78.6 to 98.5 in the study population (Table 3).

[Figure omitted. See PDF.]

Fig 2. Comparison of the level of anti-H. pylori IgG antibodies between H. pylori negative and positive samples.

This figure compares the rates of anti-H. pylori IgG antibodies between H. pylori negative and positive patients in the training set whose status had been established by standard invasive tests (i.e., histology, culture, and rapid urease test). Hence, medians of antibodies rates with interquartile intervals (IQR) were 45.0 (IQR: 75.2) and 11.9 (IQR: 7.3) U/mL respectively in H. pylori positive and negative patients. The difference of anti-H. pylori IgG levels between the two groups of patients was statistically significant (p = <0.001).

https://doi.org/10.1371/journal.pntd.0009763.g002

[Figure omitted. See PDF.]

Fig 3. ROC/AUC and best cut-of-value of the HpAfr-ELISA assay in the training set.

The plot displayed on this figure represent the ROC (receiver operating characteristic) and the AUC (Area under the ROC Curve) obtained with HpAfr-ELISA on the training set using the H. pylori status by standard invasive tests (i.e., histology, culture, and rapid urease test) as reference. This plot indicates that the ROC curve of HpAfr-ELISA had an AUC of 97.6% with its best cut-of-value being 20.2 U/mL of anti-H. pylori IgG.

https://doi.org/10.1371/journal.pntd.0009763.g003

[Figure omitted. See PDF.]

Fig 4. Sensitivity and specificity of the HpAfr-ELISA assay in the training set.

This plot indicates that the HpAfr-ELISA assay had a sensitivity of 96.7% (dashed blue line) and a specificity of 90.0% (dashed red line) when considering its best cut-off-value (dashed black line) defined at the Youden index of the dataset (the point with the maximum specificity + sensitivity).

https://doi.org/10.1371/journal.pntd.0009763.g004

[Figure omitted. See PDF.]

Fig 5. Comparison of ROCs/AUCs of the HpAfr-ELISA assay at three repeated testing.

Plots displayed on the figure represent ROC curves and AUCs obtained at repeated tests (test 1, test 2, and test 3) on the training set in intra-observer. No significant difference between the ROC curve was noted (p>0.05).

https://doi.org/10.1371/journal.pntd.0009763.g005

[Figure omitted. See PDF.]

Table 1. Qualitative agreement of HpAfr-ELISA outcomes at two repeated testing.

https://doi.org/10.1371/journal.pntd.0009763.t001

[Figure omitted. See PDF.]

Table 2. Quantitative agreement of HpAfr-ELISA outcomes at two repeated testing.

https://doi.org/10.1371/journal.pntd.0009763.t002

[Figure omitted. See PDF.]

Table 3. Bayesian estimation of the latent diagnostic performance of HpAfr-ELISA in the absence of a perfect gold standard test.

https://doi.org/10.1371/journal.pntd.0009763.t003

Epidemiology of the H. pylori infection in the study population

The community-based survey included 425 participants with a median age 29 years old (IQR: 25) and comprising 59.8% of female subjects. Anti-H. pylori IgG concentrations ranged from 3.4 to 463.7 U/mL and in total, 241 out of 425 participants (56.7%) were H. pylori-seropositive based on the cut-off-value of 20.2 U/mL. Considering prior information related to the ranges of sensitivity and specificity obtained from the training experiment (Table 3), the true prevalence of H. pylori infection in the study population was estimated 53.8% with 95%CrI lying between 42.9 and 62.7% (Table 4). Baseline characteristics of study subjects indicated that H. pylori-seropositive participants were older than seronegative ones (35.5 (IQR: 36) and 26 (IQR: 16) years old; p<0.001). In addition, H. pylori-seropositive participants comprised significantly higher proportions of individuals who had been married than single (p = 0.043), who were holding a regular profession (p = 0.003), or who were recruited from overcrowded households (p<0.001). Furthermore, a significantly smaller proportion of individuals who had an optimal level of hand hygiene (i.e., reportedly washing hands “always” after being to toilets) were H. pylori-seropositive compared to those who had non-optimal hand hygiene (i.e., who were washing hands “occasionally” to “usually”) (3.3% and 96.7%; p<0.001) (Table 5).

[Figure omitted. See PDF.]

Table 4. Bayesian estimation of the latent true prevalence of the H. pylori infection and the dyspeptic syndrome in the study population.

https://doi.org/10.1371/journal.pntd.0009763.t004

[Figure omitted. See PDF.]

Table 5. Baseline socio-demographic characteristics of study participants (n = 425).

https://doi.org/10.1371/journal.pntd.0009763.t005

All participants were interviewed for dyspepsia symptoms and their severity over a period of 6 months preceding the survey using questionnaires set for the evaluation of the DSSI score [26], the GDSS [27], and GOS Scale [28]. The feeling of stomach burn was the most frequently reported symptom (212 out of 425, 49.9%). About 56.0% of subjects reported at least one dyspepsia symptom. The prevalence of the dyspepsia syndrome as defined by the Rome IV criteria was estimated 52.2% [95%CI: 47.4; 57.1]. Dyspepsia symptoms were reportedly occurring almost daily in about 25.2% of participants. In 26.1% of cases, the dyspepsia had been interfering with usual social activities, leading to a loss of one to seven days of activities in 15.8% of cases. Though a dyspepsia syndrome had been evoked by a health care worker in 56.0% of subjects, no one had ever been specifically tested or treated for H. pylori infection. Moreover, almost all the respondents (98.6%) were unaware of H. pylori. The proportion of participants with different dyspepsia symptoms was not significantly different between H. pylori-seropositive and -seronegative subjects (p>0.05) (Tables 6 and 7).

[Figure omitted. See PDF.]

Table 6. Baseline clinical characteristics of study participants (n = 425).

https://doi.org/10.1371/journal.pntd.0009763.t006

[Figure omitted. See PDF.]

Table 7. Characterization of the dyspepsia in study participants.

https://doi.org/10.1371/journal.pntd.0009763.t007

The final logistic regression model indicated that the age of participants, crowding of households, and quality of hand hygiene were independently associated with the H. pylori-seropositivity (Table 8). Hence, for every year increase in the age of participants, the likelihood of H. pylori-seropositivity increased by one unit (adjusted odds ratio [aOR] = 1.0 [95%CI: 1.0; 1.1]; p<0.001). Subjects living in overcrowded households (i.e., where more than two residents share the same sleeping room) were more than three times associated with an H. pylori-seropositive status compared to those living in less crowded households (aOR = 3.2 [95%CI: 2.0; 5.1]; p<0.001). Individuals with a poor hand hygiene were four times and half more H. pylori-seropositive than those with good hand hygiene (aOR = 4.5 [95%CI: 2.0; 11.4]; p = 0.001). No other putative predictor or clinical feature (dyspeptic symptoms and their severity score) was significantly associated to the H. pylori-seropositive status in the model.

[Figure omitted. See PDF.]

Table 8. Logistic regression models detecting factors that predict the H. pylori seropositivity in the study population*.

https://doi.org/10.1371/journal.pntd.0009763.t008

Discussion

To our knowledge, this is the first study that develops an ELISA specifically tailored for mass screening of H. pylori infection in Africa and which is then applied to a real-world epidemiological survey conducted in African setting. This indirect ELISA system, termed HpAfr-ELISA, was conceived for the detection of anti-H. pylori IgG antibodies directed against whole-cell antigens of strains from hpAfrica1 and hpNEAfrica, the two main genetic populations of H. pylori species that infect Africans [21,31,32]. Initially, we assessed the diagnostic performance and the robustness of the assay by using a training set of samples previously collected from patients who had been tested with traditional invasive methods (i.e., histology, bacterial culture, and rapid urease test) and whose H. pylori infection status was known a priori. Then, we performed a cross-validation of this performance through samples from a community-based survey. Hence, the novel assay indicated high diagnostic performance on both the training set and the cross-validation set with sensitivity and specificity being >80%. Furthermore, the system provided consistent results in intra-observer for both qualitative (i.e., detecting H. pylori-seropositive or -seronegative status) and quantitative (i.e., anti-H. pylori IgG levels) outcomes. The performance of the HpAfr-ELISA was similar to that reported with assays tailored for other specific populations [50–55]. The needs for adjusting ELISA for H. pylori to specific populations has been raised by consistent evidence indicating that commercial ELISA kits with excellent performance in populations where they are originated from, demonstrate poorer performance when used in other regions [6,9–13]. In Africa, a survey using a commercial ELISA kit optimized at high performance in the United States (US) reported a very low sensitivity (50%) in comparison to invasive tests (e.g., culture, histology, and rapid urease test) in Egypt [56]. Inconsistent outcomes noted with ELISA kits throughout different regions likely indicate varied immunogenic properties in different population which are possibly sustained by genetic variations of H. pylori from parts of the world used as a source of antigen for the kits [9–13,57]. To adapt ELISA systems to a specific regions, local H. pylori strains could be used for preparing antigens rather than foreign strains [58]. Alternatively, common antigens shared globally by strains or strains pooled from different regions could be used [57,58]. Furthermore, best cut-off-values of antibody titers should be established and validated based on local epidemiology rather than values recommended by the assay’s manufacturer [58]. In terms of antigens, whole-cell antigens covering a broad range of bacterial antigens would provide more immunogenicity than isolated antigenic components [17]. The assay developed in the current study not only used local strains but also pooled strains from most representative genetic populations of Africa and applied whole-cell antigens rather than a specific single antigen. This may explain the high performance reached with the new assay. These results indicate that the HpAfr-ELISA assay is probably a useful tool for assessing accurately the H. pylori infection in African setting, notably in the DRC. Obviously, H. pylori serological tests detect individual who had been exposed to the pathogen including past infections cases. Consequently, they probably can be more useful as screening tools for populations than as diagnostic tools for individual diagnosis. As such, during strategies for the management of H. pylori like the “test-and-treat” approach [2,59], these tests would be helpful for filtering the vast majority of uninfected patients to allow focusing available resources on the most likely infected cases. Moreover, serological tests with their relative low costs and high-throughput capabilities can also be useful for evaluating and monitoring the disease burden in populations over periods of time. To further illustrate the usefulness of HpAfr-ELISA, we applied the assay for estimating the true prevalence of H. pylori infection and for identifying factors that expose to H. pylori in a community from the DRC.

Estimating the true prevalence of a disease constitute an important part of public health efforts that can help define efficient health policies and optimally allocating available resources. Bayesian methods applied in this study make it possible to estimate accurately the true prevalence in the absence of a perfect gold standard test [42,43,60]. The true prevalence of H. pylori in the study population was thus estimated 53.8% with a 95%CrI ranging from 42.8 to 62.7%. At the time of this publication, there was no study reporting the prevalence of H. pylori infection at community level within the DRC. Although consistently supporting a high prevalence of H. pylori in DRC, previous reports had only involved patients recruited in hospitals and are likely not faithfully reflecting the epidemiological situation in the general population of respective study areas [61–66]. Likewise, other observations also suggest a high prevalence in neighboring countries [67–70]. Furthermore, recent global meta-analyses of community-based studies indicated that such epidemiological situation might be present in most emerging and developing countries, especially in Africa, in contrast to developed countries where the H. pylori prevalence could be lower [3,4]. The high prevalence of H. pylori probably testify a persistent bacterial transmission coupled with limited access to adequate diagnosis and treatment of the infection in the studied community [71]. It is likely that the majority of seropositive subjects had chronic H. pylori infection, a condition associated with severe complications such as peptic ulcers and gastric cancers [2,72]. We identified an increased age as significantly predicting the H. pylori-seropositive status in this study. The link between H. pylori and age is widely discussed in previous studies [73–75]. Globally, in developed countries where the H. pylori prevalence seems to be decreasing during the last two decades, the link with age is likely due to the effect of birth cohorts of earlier generations that had been more exposed in the past to favoring factors such as poor sanitation [76]. In developing countries, oppositely, the prevalence of H. pylori infection seems to have remained stable over years [4]. Thus, the increase of H. pylori infections with growing age in these countries might result from an accumulation of infected but untreated people over time and in conditions persistently promoting the infection transmission. Long-term follow-up studies are however warranted to evaluate this hypothesis. Given the potential accumulation of infections with age, elders should attract maximum attention of health policy makers as they are additionally the most prone to severe H. pylori-related complications such as gastric cancers and peptic ulcer diseases with significantly higher mortality [76,77]. Beside the effect of age, life in overcrowded households and poor hand hygiene were also associated with the H. pylori-seropositive status. This suggests that conditions with low hygiene and precarious sanitation might be favoring H. pylori infections or perpetuating the human-to-human transmission of the bacteria in the study population. These observations are consistent with the global consensus on the epidemiology and the most likely transmission routes of H. pylori (e.g., oro-fecal and oro-oral routes) [78–80]. Therefore, enhanced efforts for improving individual hygiene are required to reduce the burden of H. pylori infection locally. Investigations on the main transmission routes and the age of acquisition should be undergone for strengthening the basis for effective prevention strategies.

This study comprised several limitations. The community-based survey included only adults from a specific location in the country. In addition, even though the serological test was created using strains of genetic populations known as colonizing most of the African continent, it was only tested against samples from DRC. Strains used for preparing the assay did not include those from hpAfrica2, a very divergent H. pylori population colonizing individuals from a specific ethnic of hunter-gatherers in southern Africa known as the San population [21,31,32]. Similarly, the assay did not take into account bacterial populations that had been introduced to Africa in recent history (e.g., hpEurope and hpAsia2) [20,21]. The generalizability of outcomes would thus need to be interpreted with caution. This survey was based on interviews and cannot be spared from memory bias that this entails. As an indirect diagnostic method, the assay developed in this study is prone to performance errors and misclassifications of samples. To reduce the possibility of these errors, we designed the assay to account for the local antigenicity background. We additionally used Bayesian methods to control statistical uncertainties in observations and to provide more accurate estimates despite the absence of a perfect gold standard test [42,43,60]. Though the validity of the assay could be shown, the assessment processed only in intra-observer without any external validation. Furthermore, the possibility of cross-reactivity between the assay and heterologous species or other host-related pathophysiological conditions could not be investigated. The assay remains thus perfectible and open to further improvements. The seropositivity to the HpAfr-ELISA assay did not correlate significantly with symptoms in this study. Though this is a general issue for serological tests of H. pylori infection, it can limit their utility for individual diagnosis [81]. Further evidence on the usefulness and the cost-effectiveness of this new kit should be provided in future studies that should include a comparative analysis against commercially available ELISA kits or other non-invasive tests of H. pylori (e.g., stool antigen test and UBT). The logistic regression models fitted to identify factors associated with the exposition to H. pylori in the community did not take the test sensitivity and specificity into account, but we acknowledge that the practical feasibility of this is not as easy as it seems. Finally, despite the high performance achieved by the assay, the Bayesian methods applied to assess the assay tool are still less commonly used in the medical field and this unfamiliarity may be seen also as a limitation for this study.

Notwithstanding the above limitations, this study has the merit of having successfully developed and validated the first ELISA specifically tailored for screening H. pylori infection in Africa. We believe that this assay is robust and provides reproducible outcomes with high diagnostic performance. Bayesian modeling enabled estimating a high latent true prevalence of H. pylori infection in the studied population that requires being addressed with appropriate public health strategies. Factors likely favoring H. pylori infection in the study area, i.e., growing age, life in overcrowded households, and poor hand hygiene, should specifically be targeted by management and mitigation strategies against the infection. Future explorations are warranted to further improve the novel ELISA system and to make it accessible for the needs of practical use. Operational research should be further implemented to extend our understanding of the epidemiology of H. pylori and to address this prevalent infection programmatically and efficiently in the study area and globally in Africa.

Supporting information

S1 Fig. Neighbor-joining phylogenetic tree inferring the genetic populations of H. pylori strains used for the HpAfr-ELISA along with reference strains.

https://doi.org/10.1371/journal.pntd.0009763.s001

(PDF)

S2 Fig. Electrophoretic profile of Whole-cell antigen proteins used for the HpAfr-ELISA.

https://doi.org/10.1371/journal.pntd.0009763.s002

(PDF)

S1 Table. STROBE Statement–Checklist of items that are included in the reports.

https://doi.org/10.1371/journal.pntd.0009763.s003

(PDF)

S2 Table. Characteristics of the training samples used for the HpAfr-ELISA system (n = 60).

https://doi.org/10.1371/journal.pntd.0009763.s004

(PDF)

S1 File. Bayesian statistical modeling.

https://doi.org/10.1371/journal.pntd.0009763.s005

(PDF)

Acknowledgments

We acknowledge the technical support given by Dr. Edouard Kanda, Dr. Prisca Musau, Dr. Valencia Tshidibi, Dr. Larissa Nsumpi, Dr. Barthelemy Cyebwe, and Dr. Denis Mbuyi, who collected DBS samples and conducted interviews on the field.

Citation: Tshibangu-Kabamba E, Phuc BH, Tuan VP, Fauzia KA, Kabongo-Tshibaka A, Kayiba NK, et al. (2021) Assessment of the diagnostic accuracy and relevance of a novel ELISA system developed for seroepidemiologic surveys of Helicobacter pylori infection in African settings. PLoS Negl Trop Dis 15(9): e0009763. https://doi.org/10.1371/journal.pntd.0009763

References

1. Yamaoka Y. Mechanisms of disease: Helicobacter pylori virulence factors. Nature reviews Gastroenterology & hepatology. 2010;7(11):629. pmid:20938460

2. Yamaoka Y. How to eliminate gastric cancer-related death worldwide? Nature Reviews Clinical Oncology. 2018;15(7):407–8. pmid:29700378

3. Zamani M, Ebrahimtabar F, Zamani V, Miller W, Alizadeh-Navaei R, Shokri-Shirvani J, et al. Systematic review with meta-analysis: the worldwide prevalence of Helicobacter pylori infection. Alimentary pharmacology & therapeutics. 2018;47(7):868–76. pmid:29430669

4. Hooi JKY, Lai WY, Ng WK, Suen MMY, Underwood FE, Tanyingoh D, et al. Global Prevalence of Helicobacter pylori Infection: Systematic Review and Meta-Analysis. Gastroenterology. 2017;153(2):420–9. Epub 2017/05/01. pmid:28456631.

5. Wang Y-K, Kuo F-C, Liu C-J, Wu M-C, Shih H-Y, Wang SS, et al. Diagnosis of Helicobacter pylori infection: Current options and developments. World Journal of Gastroenterology: WJG. 2015;21(40):11221. pmid:26523098

6. Patel SK, Pratap CB, Jain AK, Gulati AK, Nath G. Diagnosis of Helicobacter pylori: what should be the gold standard? World journal of gastroenterology. 2014;20(36):12847–59. Epub 2014/10/04. pmid:25278682; PubMed Central PMCID: PMC4177467.

7. Miftahussurur M, Yamaoka Y. Diagnostic Methods of Helicobacter pylori Infection for Epidemiological Studies: Critical Importance of Indirect Test Validation. BioMed research international. 2016;2016:4819423. Epub 2016/02/24. pmid:26904678; PubMed Central PMCID: PMC4745376.

8. Miftahussurur M, Yamaoka Y. Diagnostic Methods of Helicobacter pylori Infection for Epidemiological Studies: Critical Importance of Indirect Test Validation. Biomed Res Int. 2016;2016:4819423–. Epub 01/19. pmid:26904678.

9. Burucoa C, Delchier JC, Courillon-Mallet A, de Korwin JD, Mégraud F, Zerbib F, et al. Comparative evaluation of 29 commercial Helicobacter pylori serological kits. Helicobacter. 2013;18(3):169–79. Epub 2013/01/16. pmid:23316886.

10. Hoang TTH, Wheeldon T-U, Bengtsson C, Phung DC, Sörberg M, Granström M. Enzyme-linked immunosorbent assay for Helicobacter pylori needs adjustment for the population investigated. J Clin Microbiol. 2004;42(2):627–30. pmid:14766827.

11. Mohammadi M, Talebkhan Y, Khalili G, Mahboudi F, Massarrat S, Zamaninia L, et al. Advantage of using a home-made ELISA kit for detection of Helicobacter pylori infection over commercially imported kits. Indian journal of medical microbiology. 2008;26(2):127–31. Epub 2008/05/01. pmid:18445947.

12. Thong-Ngam D, Chayanupatkul M, Vongchampa P, Hanvivatvong O. An evaluation of a new in-house serum and urine ELISA test for detection of Helicobacter pylori infection in Thai population. J Med Assoc Thai. 2011;94(8):985–90. pmid:21863682.

13. Leung WK, Ng EK, Chan FK, Chung SC, Sung JJ. Evaluation of three commercial enzyme-linked immunosorbent assay kits for diagnosis of Helicobacter pylori in Chinese patients. Diagnostic microbiology and infectious disease. 1999;34(1):13–7. Epub 1999/05/26. pmid:10342102.

14. Best LM, Veldhuyzen van Zanten SJ, Bezanson GS, Haldane DJ, Malatjalian DA. Serological detection of Helicobacter pylori by a flow microsphere immunofluorescence assay. J Clin Microbiol. 1992;30(9):2311–7. pmid:1400995.

15. Kronsteiner B, Bassaganya-Riera J, Philipson C, Viladomiu M, Carbo A, Abedi V, et al. Systems-wide analyses of mucosal immune responses to Helicobacter pylori at the interface between pathogenicity and symbiosis. Gut Microbes. 2016;7(1):3–21. pmid:26939848.

16. Guo L, Yang H, Tang F, Yin R, Liu H, Gong X, et al. Oral Immunization with a Multivalent Epitope-Based Vaccine, Based on NAP, Urease, HSP60, and HpaA, Provides Therapeutic Effect on H. pylori Infection in Mongolian gerbils. Front Cell Infect Microbiol. 2017;7:349–. pmid:28824883.

17. Mahdi BM. Immune response to Helicobacter pylori. Trends in Helicobacter pylori Infection. 2014:79.

18. Vitoriano I, Rocha-Gonçalves A, Carvalho T, Oleastro M, Calado CR, Roxo-Rosa M. Antigenic diversity among Portuguese clinical isolates of Helicobacter pylori. Helicobacter. 2011;16(2):153–68. Epub 2011/03/26. pmid:21435094.

19. Smith S, Fowora M, Pellicano R. Infections with Helicobacter pylori and challenges encountered in Africa. World journal of gastroenterology. 2019;25(25):3183–95. Epub 2019/07/25. pmid:31333310; PubMed Central PMCID: PMC6626727.

20. Montano V, Didelot X, Foll M, Linz B, Reinhardt R, Suerbaum S, et al. Worldwide population structure, long-term demography, and local adaptation of Helicobacter pylori. Genetics. 2015;200(3):947–63. pmid:25995212

21. Moodley Y, Linz B, Bond RP, Nieuwoudt M, Soodyall H, Schlebusch CM, et al. Age of the association between Helicobacter pylori and man. PLoS pathogens. 2012;8(5):e1002693. Epub 2012/05/17. pmid:22589724; PubMed Central PMCID: PMC3349757.

22. Gluud C, Gluud LL. Evidence based diagnostics. BMJ. 2005;330(7493):724–6. pmid:15790646

23. Von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. International journal of surgery. 2014;12(12):1495–9. pmid:25046131

24. Tshibangu-Kabamba E, Ngoma-Kisoko PdJ, Tuan VP, Matsumoto T, Akada J, Kido Y, et al. Next-Generation Sequencing of the Whole Bacterial Genome for Tracking Molecular Insight into the Broad-Spectrum Antimicrobial Resistance of Helicobacter pylori Clinical Isolates from the Democratic Republic of Congo. Microorganisms. 2020;8(6):887. pmid:32545318

25. Obuchowski NA, McClish DK. Sample size determination for diagnostic accuracy studies involving binormal ROC curve indices. Statistics in medicine. 1997;16(13):1529–42. Epub 1997/07/15. pmid:9249923.

26. Leidy NK, Farup C, Rentz AM, Ganoczy D, Koch KL. Patient-based assessment in dyspepsia: development and validation of Dyspepsia Symptom Severity Index (DSSI). Digestive diseases and sciences. 2000;45(6):1172–9. Epub 2000/07/06. pmid:10877234.

27. el-Omar EM, Banerjee S, Wirz A, McColl KE. The Glasgow Dyspepsia Severity Score—a tool for the global measurement of dyspepsia. European journal of gastroenterology & hepatology. 1996;8(10):967–71. Epub 1996/10/01. pmid:8930560.

28. Van Zanten SV, Chiba N, Armstrong D, Barkun A, Thomson A, Mann V, et al. Validation of a 7-point Global Overall Symptom scale to measure the severity of dyspepsia symptoms in clinical trials. Aliment Pharmacol Ther. 2006;23(4):521–9. pmid:16441473

29. Stanghellini V, Chan FK, Hasler WL, Malagelada JR, Suzuki H, Tack J, et al. Gastroduodenal disorders. Gastroenterology. 2016;150(6):1380–92. pmid:27147122

30. Grüner N, Stambouli O, Ross RS. Dried blood spots-preparing and processing for use in immunoassays and in molecular techniques. Journal of visualized experiments: JoVE. 2015;(97).

31. Nell S, Eibach D, Montano V, Maady A, Nkwescheu A, Siri J, et al. Recent acquisition of Helicobacter pylori by Baka pygmies. PLoS genetics. 2013;9(9):e1003775. pmid:24068950

32. Kabamba ET, Tuan VP, Yamaoka Y. Genetic populations and virulence factors of Helicobacter pylori. Infection, Genetics and Evolution. 2018;60:109–16. pmid:29471116

33. Shiota S, Cruz M, Abreu JAJ, Mitsui T, Terao H, Disla M, et al. Virulence genes of Helicobacter pylori in the Dominican Republic. Journal of medical microbiology. 2014;63(Pt 9):1189. pmid:24965801

34. Mwangi C, Njoroge S, Tshibangu-Kabamba E, Moloo Z, Rajula A, Devani S, et al. Whole Genome Sequencing Reveals Virulence Potentials of Helicobacter pylori Strain KE21 Isolated from a Kenyan Patient with Gastric Signet Ring Cell Carcinoma. Toxins. 2020;12(9):556. pmid:32872465

35. Team RC. R: A language and environment for statistical computing. 2013.

36. Wickham H, Wickham MH. Package tidyverse. Easily Install and Load the ‘Tidyverse. 2017.

37. Lê S, Josse J, Husson F. FactoMineR: an R package for multivariate analysis. Journal of Statistical Software. 2008;25(1):1–18.

38. Kassambara A, Mundt F. Factoextra: extract and visualize the results of multivariate data analyses. R package version 1.0. 4. 2017. 2017.

39. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez J-C, et al. Package ‘pROC’. 2019.

40. Revelle W. An overview of the psych package. Department of Psychology Northwestern University Accessed on March. 2011;3(2012):1–25.

41. Gouhier TC, Guichard F. Synchrony: quantifying variability in space and time. Methods in Ecology and Evolution. 2014;5(6):524–33.

42. Speybroeck N, Devleesschauwer B, Joseph L, Berkvens D. Misclassification errors in prevalence estimation: Bayesian handling with care. International journal of public health. 2013;58(5):791–5. Epub 2012/12/25. pmid:23263198.

43. Branscum AJ, Gardner IA, Johnson WO. Estimation of diagnostic-test sensitivity and specificity through Bayesian modeling. Preventive veterinary medicine. 2005;68(2–4):145–63. Epub 2005/04/12. pmid:15820113.

44. Devleesschauwer B, Torgerson PR, Charlier J, Levecke B, Praet N, Dorny P, et al. Package ‘prevalence’. 2013.

45. Plummer M. JAGS: Just another Gibbs sampler. 2004.

46. Plummer M, Best N, Cowles K, Vines K. Package ‘coda’. 2019.

47. Plummer M, Best N, Cowles K, Vines K. CODA: convergence diagnosis and output analysis for MCMC. R news. 2006;6(1):7–11.

48. Gallagher K, Charvin K, Nielsen S, Sambridge M, Stephenson J. Markov chain Monte Carlo (MCMC) sampling methods to determine optimal models, model resolution and model choice for Earth Science problems. Marine and Petroleum Geology. 2009;26(4):525–35.

49. Gelman A, Rubin DB. Inference from iterative simulation using multiple sequences. Statistical science. 1992;7(4):457–72.

50. Aziz F, Sherwani SK, Akhtar SS, Kazmi SU. Development of an in-house enzyme-linked immunosorbent assay based on surface whole cell antigen for diagnosis of Helicobacter pylori infection in patients with gastroduodenal ulcer disease. World Journal of Microbiology and Biotechnology. 2014;30(1):305–15. pmid:23921679

51. Ogata SK, Camorlinga-Ponce M, Granato CFH, Rohr MRdS, Artigiani Neto R, Kawakami E. Development and validation of a whole-cell ELISA for serologically diagnosing Helicobacter pylori infection in Brazilian children and adults: a diagnostic accuracy study. Sao Paulo Medical Journal. 2018;136(5):442–8. pmid:30570095

52. Lee S-Y, Moon H-W, Hur M, Yun Y-M. Validation of western Helicobacter pylori IgG antibody assays in Korean adults. Journal of Medical Microbiology. 2015;64(5):513–8. pmid:25752852

53. Shimoyama T, Sawada Y, Sawada N, Chinda D, Fukuda S. Accuracy of stick-type kit and enzyme-linked immunosorbent assay for the detection of urinary Helicobacter pylori antibodies in a population of Japan Sea region of northern Japan. Japanese journal of infectious diseases. 2016:JJID. 2015.642.

54. Matsuo Y, Kido Y, Akada J, Shiota S, Binh TT, Trang TTH, et al. Novel CagA ELISA exhibits enhanced sensitivity of Helicobacter pylori CagA antibody. World journal of gastroenterology. 2017;23(1):48. pmid:28104980

55. Kodama M, Okimoto T, Mizukami K, Fukuda K, Ogawa R, Okamoto K, et al. Evaluation of a Novel Anti-H. pylori Antibody Detection Kit by Latex Turbidimetric Immunoassay. Clinical laboratory. 2019;65(6). pmid:31232019

56. Frenck RW, Fathy HM, Sherif M, Mohran Z, El Mohammedy H, Francis W, et al. Sensitivity and specificity of various tests for the diagnosis of Helicobacter pylori in Egyptian children. Pediatrics. 2006;118(4):e1195–e202. pmid:16982805

57. Höök-Nikanne J, Perez-Perez GI, Blaser MJ. Antigenic characterization of Helicobacter pylori strains from different parts of the world. Clinical and diagnostic laboratory immunology. 1997;4(5):592–7. pmid:9302211

58. Marchildon PA, Sugiyama T, Fukada Y, Peacock JS, Asaka M, Shimoyama T, et al. Evaluation of the effects of strain-specific antigen variation on the accuracy of serologic diagnosis of Helicobacter pylori infection. J Clin Microbiol. 2003;41(4):1480–5. pmid:12682133

59. Gisbert JP, Calvet X. Helicobacter pylori “test-and-treat” strategy for management of dyspepsia: a comprehensive review. Clinical and translational gastroenterology. 2013;4(3):e32. pmid:23535826

60. Lewis FI, Torgerson PR. A tutorial in estimating the prevalence of disease in humans and animals in the absence of a gold standard diagnostic. Emerging themes in epidemiology. 2012;9(1):9. Epub 2012/12/29. pmid:23270542; PubMed Central PMCID: PMC3558341.

61. Longo-Mbenza B, Apalata T, Longokolo M, Mambimbi MM, Mokondjimobe E, Gombet T, et al. Association of Helicobacter pylori infection with the metabolic syndrome among HIV-infected black Africans receiving highly active antiretroviral therapy. Cardiovascular journal of Africa. 2015;26(2):52–6. Epub 2015/05/06. pmid:25940117; PubMed Central PMCID: PMC4815505.

62. On’kin J, Longo-Mbenza B, Tchokonte-Nana V, Okwe AN, Kabangu NK. Hyperbolic relation between beta-cell function and insulin sensitivity for type 2 diabetes mellitus, malaria, influenza, Helicobacter pylori, Chlamydia pneumoniae, and hepatitis C virus infection-induced inflammation/oxidative stress and temporary insulin resistance in Central Africans. Turkish journal of medical sciences. 2017;47(6):1834–41. Epub 2018/01/08. pmid:29306246.

63. Longo-Mbenza B, Nkondi Nsenga J, Vangu Ngoma D. Prevention of the metabolic syndrome insulin resistance and the atherosclerotic diseases in Africans infected by Helicobacter pylori infection and treated by antibiotics. International journal of cardiology. 2007;121(3):229–38. Epub 2007/03/21. pmid:17368586.

64. Longo-Mbenza B, Mambune HF, Kasiam JB, Vita EK, Fuele SM, Nsenga JN, et al. Relationship between waist circumference and cholesterol in Central Africans with congestive heart failure. West African journal of medicine. 2007;26(3):183–90. Epub 2008/04/11. pmid:18399331.

65. Glupczynski Y, Bourdeaux L, Verhas M, Deprez C, DeVos D, Devreker T. Use of a urea breath test versus invasive methods to determine the prevalence of Helicobacter pylori in zaire. European Journal of Clinical Microbiology and Infectious Diseases. 1992;11(4):322–7. pmid:1396751

66. Nsenga JN, Longo-Mbenza B, Mpolesha J-MK, Mabwa L, Mbungu S, Vangu D. Séropositivité pour Helicobacter pylori, gastrite à Helicobacter pylori, composantes de la pression artérielle: réalité du risque cardiovasculaire. Journal Africain d’Hépato-Gastroentérologie. 2007;1(2):87–92.

67. Gupta V, Perez-Perez GI, Dorsey G, Rosenthal PJ, Blaser MJ. The seroprevalence of Helicobacter pylori and its relationship to malaria in Ugandan children. Transactions of the Royal Society of Tropical Medicine and Hygiene. 2012;106(1):35–42. pmid:22018600

68. Hestvik E, Tylleskar T, Kaddu-Mulindwa DH, Ndeezi G, Grahnquist L, Olafsdottir E, et al. Helicobacter pylori in apparently healthy children aged 0–12 years in urban Kampala, Uganda: a community-based cross sectional survey. BMC gastroenterology. 2010;10(1):62. pmid:20553588

69. Mbulaiteye SM, Gold BD, Pfeiffer RM, Brubaker GR, Shao J, Biggar RJ, et al. H. pylori-infection and antibody immune response in a rural Tanzanian population. Infectious agents and cancer. 2006;1(1):3.

70. McLaughlin N, McLaughlin D, Lefcort H. The influence of socio-economic factors on Helicobacter pylori infection rates of students in rural Zambia. 2003.

71. Hassan C, Aabakken L, Ebigbo A, Karstensen JG, Guy C, Dinis-Ribeiro M, et al. Partnership with African Countries: European Society of Gastrointestinal Endoscopy (ESGE)–Position Statement. Endoscopy international open. 2018;6(10):E1247–E55. pmid:30302382

72. Sugano K, Tack J, Kuipers EJ, Graham DY, El-Omar EM, Miura S, et al. Kyoto global consensus report on Helicobacter pylori gastritis. Gut. 2015;64(9):1353–67. pmid:26187502

73. Ahmed K, Khan A, Ahmed I, Tiwari S, Habeeb A, Ahi J, et al. Impact of household hygiene and water source on the prevalence and transmission of Helicobacter pylori: a South Indian perspective. Singapore medical journal. 2007;48(6):543. pmid:17538754

74. Alizadeh A, Ansari S, Ranjbar M, Shalmani H, Habibi I, Firouzi M, et al. Seroprevalence of Helicobacter pylori in Nahavand: a population-based study. 2009.

75. Ameri G, Alkadasi M. The prevalence of Helicobacter pylori and risk factors infection a ssociated in Taiz city, Yemen. Int J Curr Microbiol App Sci. 2013;2:226–33.

76. Pilotto A, Franceschi M. Helicobacter pylori infection in older people. World Journal of Gastroenterology: WJG. 2014;20(21):6364. pmid:24914358

77. Pilotto A, Salles N. Helicobacter pylori infection in geriatrics. Helicobacter. 2002;7:56–62. pmid:12197911

78. Malaty H, Graham D, Klein D, Adam EE, Evans D. Transmission of Helicobacter pylori infection studies in families of healthy individuals. Scandinavian journal of gastroenterology. 1991;26(9):927–32. pmid:1947784

79. Mladenova I, Durazzo M. Transmission of Helicobacter pylori. Minerva gastroenterologica e dietologica. 2018;64(3):251–4. pmid:29458239

80. Brown LM. Helicobacter pylori: epidemiology and routes of transmission. Epidemiologic reviews. 2000;22(2):283–97. pmid:11218379

81. Akada J, Tshibangu-Kabamba E, Tuan VP, Kurogi S, Matsuo Y, Ansari S, et al. Serum Helicobacter pylori antibody reactivity in seven Asian countries using an automated latex aggregation turbidity assay. Journal of Gastroenterology and Hepatology. 2021. pmid:33609333