HIV-Helicobacter pylori Co-Infection: Antibiotic Resistance, Prevalence, and Risk Factors

Background Patients infected with human immunodeficiency virus (HIV) are living longer due to the availability of more potent treatments. However, prescription of antibiotics to treat or prevent infections in these patients may increase the likelihood of co-infection with antibiotic-resistant species. Aim To compare antimicrobial susceptibility of Helicobacter pylori (H. pylori) in HIV-positive and HIV-negative patients and assess risk-factors for resistance. Methods We prospectively collected data from consecutive HIV-positive and HIV-negative patients undergoing upper gastrointestinal endoscopy. Patients with H. pylori-positive gastric biopsies who had never received H. pylori treatment were included. Results Of the 353 patients included, 93 were HIV-positive and 260 HIV-negative. Among the HIV-positive patients, 56 (60%) had been infected for <10 years, the median CD4+ count was 493 cells/μl and median viral load was 61 copies/mL; 66 (71%) were receiving antiretroviral therapy. HIV-positive patients were more often male (p = 0.009), had a lower body mass index (p<0.0001), and had less frequently received antibiotics during the 12-months prior to the endoscopy (p<0.0001) than HIV-negative patients. HIV-positive patients were more likely to have H. pylori resistant to levofloxacin (p = 0.0004), metronidazole (p = 0.01), or multiple antibiotics (p = 0.006). HIV-positive Black Africans were more likely to have resistant strains than were HIV-negative Black Africans (p = 0.04). Ethnicity and HIV status were independent risk factors for H. pylori resistance in all patients and acquired immune deficiency syndrome (AIDS) and sex were risk factors in HIV-positive patients. Conclusions There was a higher prevalence of primary H. pylori-resistant strains in HIV-positive than in HIV-negative patients. AIDS and sex were predictors of H. pylori resistance in HIV-positive patients.


Introduction
Human immunodeficiency virus (HIV) and Helicobacter pylori (H. pylori) infections are worldwide healthcare burdens. Improvements in antiretroviral treatment and patient care have increased the life expectancy of HIV-positive individuals [1]. As a consequence, HIV-positive patients suffering from H. pylori infection are seen more frequently than was the case before the advent of highly active antiretroviral therapy (HAART). We previously demonstrated increased H. pylori infection in HIV-positive patients who underwent upper gastrointestinal (UGI) endoscopy after the introduction of HAART compared to those in the pre-HAART era [2]. The prevalence of H. pylori infection among HIV-positive patients is variable, ranging from 10% to 76%, depending on the population studied and the geographic area [3,[4][5][6][7]. In a recent study in our center, the prevalence of H. pylori co-infection was 48/145 (33%) among symptomatic HIV-positive patients who underwent UGI endoscopy, and was more common in patients of Black African origin and in heterosexuals [3].
Complications related to H. pylori infection in the general population, such as gastro-duodenal ulcer and gastric carcinoma, have been extensively documented in different geographic regions. To avoid such complications, treatment to eradicate H. pylori is recommended [8]. Different therapeutic regimens have been recommended. The first-line treatment worldwide is triple therapy with a proton pump inhibitor (PPI) plus two antibiotics. In the general population, the success of triple therapy ranges from 85% to 90%. However, an increasing incidence of H. pylori treatment failure has been reported [9][10][11]. This has been attributed to various factors including patient compliance, polymorphism for PPI metabolism, and H. pylori susceptibility, which is the main factor affecting treatment outcome [8][9][10][11][12][13]. Increased antibiotic resistance has been described in various regions and has been attributed to high antibiotic consumption [13][14][15]. This finding may be particularly relevant for HIV-positive patients who are often exposed to antibiotics for chemoprevention of opportunistic diseases or treatment of acute infections.
Few studies have evaluated the impact of antibiotics on HIV-H. pylori co-infected patients. We, therefore, evaluated H. pylori susceptibility among HIV-positive and-negative patients seen at our center, and looked for predictive factors for H. pylori resistance.

Patients
This longitudinal observational study was carried out at CHU Saint-Pierre in Brussels, a general hospital that currently monitors close to 3000 HIV-positive patients on a regular basis and performs more than 2700 UGI endoscopies per annum.
We prospectively collected data on consecutive HIV-positive patients undergoing UGI endoscopy between 1st January 2008 and 31 December 2011. Patients with gastric biopsy samples positive for H. pylori infection on pathology and culture, and who had not received previous H. pylori treatment were eligible for inclusion in the study. During the same period, we collected data on a control group of consecutive HIV-negative patients undergoing UGI endoscopy prior to bariatric surgery for obesity who met the same criteria for H. pylori diagnosis and were naïve to H. pylori treatment. In our hospital, H. pylori eradication is mandatory before bariatric surgery (gastric bypass) for obesity.

Methods
The study was conducted in conformity with the declaration of Helsinki and the protocol was approved by the local hospital ethics committee at CHU Saint-Pierre in Brussels. All procedures described in the study were performed for routine medical purposes. Written consent was obtained from all patients.
Parameters collected on the day of UGI endoscopy included demographics (age, sex, ethnicity, body mass index [BMI]), HIV status and parameters (duration of HIV infection, Center for Disease Control (CDC) stage, viral load, antiretroviral treatment, T-CD4 [CD4+] cell count), Toxoplasma gondii serology, closest to the date of endoscopy, and antibiotic use (including amoxicillin (AMX), clarithromycin (CLA), fluoroquinolones, tetracycline (TET), and metronidazole (MTZ)), Toxoplasma gondii or Pneumocystis carinii chemoprevention, or antimalarial drug use (including mefloquine, atovaquone, chloroquine, primaquine) within twelve months prior to endoscopy. Bismuth compounds are rarely used in Belgium, and were not recorded. Data were also collected on H. pylori treatment, including antibiotic susceptibility, type of treatment, tolerance, and response rate to first-line anti-H. pylori triple therapy.
Endoscopy and gastric sample biopsies. Patients underwent UGI endoscopy after fasting for 12 hours or overnight. At least eight gastric biopsy samples were taken (four from the angulus and four from the body). Four of the samples (two from each area) were immediately transferred to Amies Agar Gel Collection and Transport Swabs (Copan Diagnostic Inc., USA), and placed in a refrigerator at -5 to +5°C before being sent within 24 hours to the Department of Microbiology. The remaining samples were preserved in formaldehyde for histopathology.
H. pylori isolation and susceptibility testing. Upon receipt by the laboratory, gastric biopsy specimens were either frozen at -70°C until required, or processed immediately. Each specimen was homogenized in sterile water for 15 s and inoculated onto selective agar plates (in-house and Helicobacter agar Becton Dickinson, USA) by circular streaking with a bent pipette. Plates were incubated for 3 to 7 days at 37°C in a humid, microaerophilic incubator (Binder [serial no 08-51907], Germany) with a final atmosphere of 5% to 6% O 2 , 8% to 10% CO 2 , and 80% to 85% N 2 .
Pathology. Immunohistological staining was used to diagnose H. pylori infection. All slides were interpreted by the same pathologists according to the updated Sydney system scoring [18].
H. pylori treatment regimen. Depending on the antibiotic susceptibility, patients received one of the following regimens: • triple therapy: a PPI (omeprazole 20 mg, esomeprazole 40 mg, or pantoprazole 40 mg) plus two antibiotics (AMX 1000 mg, CLA 500 mg, TET 500 mg, or MTZ 500 mg) twice daily for 10 days (range 7 to 14 days) • sequential therapy: a PPI plus amoxicillin 1000 mg twice daily for 5 days followed by a PPI plus clarithromycin 500 mg and metronidazole 500 mg twice daily for 5 days • quadruple therapy: a PPI twice daily, TET-HCL 500 mg plus colloidal bismuth subcitrate 500 mg four times daily, and metronidazole 500 mg three times daily for 10 days.
Urea breath test (UBT). A standard UBT protocol was performed after overnight fasting. Briefly, an exhaled breath sample was obtained before (T0), and at 10 (T1) and 15 (T2) minutes after administration of two microcuries of 14 C-urea liquid mixed with 30 ml of water given orally. Samples T0, T1, and T2 were analyzed using a PerkinElmer Tricarb-1500 scintillation counter. The final result was obtained using the formula: [T1 minus T0, divided by the value provided by the scintillation counter for a 1/1000 dilution of 2 microcuries standard dose] multiplied by the weight (in kilograms). The T2 value was used as a control. A value <1.5 was defined as negative, indicating eradication of H. pylori infection [19].
The test was performed five to eight weeks after the end of the H. pylori treatment regimen. To insure correct interpretation of the results, patients were asked to confirm that no other antibiotic or PPI treatment had been administered since completion of H. pylori therapy.
Statistics. Patients were classified according to HIV status (positive or negative) and data from these two groups were compared. Descriptive statistics are expressed using mean values and range for quantitative measurements and percentages for qualitative measurements. Fisher's exact test and the chi-square test were used for comparisons between groups. Betweengroup differences are presented as the difference between the means of each group with the 95% confidence interval (95% CI) for each comparison. A crude odds ratio (OR) with its 95% CI is presented for each comparison.
Predictors of H. pylori resistance were assessed in all patients using the chi-square test and were confirmed by logistic regression in multivariable analysis. Parameters used for the study of predictors of resistance in this group included age, sex, ethnicity, and HIV status.
Predictors of H. pylori resistance were assessed in HIV-positive patients using the chi-square test and were confirmed by logistic regression in multivariable analysis for any parameter with a p-value 0.1 in univariate analysis. Parameters used for the study of predictors of resistance in this group included sex, ethnicity, HIV viral load and duration of HIV infection, antiretroviral treatment, and a diagnosis of acquired immunodeficiency syndrome (AIDS).
Analyses were performed using SAS statistical software (version 9Á2; SAS Institute, Cary, NC, USA). A p value of < 0.05 was considered to be significant.

Results
Of 372 patients with positive tests for H. pylori (99 HIV-positive and 273 HIV-negative), 353 (93 HIV-positive and 260 HIV-negative) were naïve to H. pylori treatment and were included in the study (Fig 1). Among the HIV-infected patients, 31 (44%) had a diagnosis of AIDS and 66 (71.0%) were receiving antiretroviral therapy. The median CD4 count was 489 cells/μl, but was lower in Black Africans and North Africans than in Caucasian Europeans (p = 0.03). Toxoplasma gondii (T. gondii) IgG antibodies were present in 43/80 (53.1%) patients screened; 20/ 93 (21%) patients were receiving trimethoprim-sulfamethoxazole for chemoprevention against T. gondii or Pneumocystosis carnii or for an acute infection.
Compared to HIV-negative patients, HIV-positive patients were more often male (p = 0.009), had a lower BMI (p<0.0001), and had less frequently received antibiotics during the 12 months prior to UGI endoscopy (p<0.0001) ( Table 1). In the HIV-positive patients, those with AIDS were more likely to have received antibiotics during the 12 months prior to  Overall, H. pylori resistance was present in 199 (56.3%) patients. Resistance to MTZ was observed most frequently, followed by resistance to LEV and then CLA ( Table 2). H. pylori resistance to LEV (p = 0.0004), MTZ (p = 0.01), or multiple antibiotics was more frequent in HIVpositive than in HIV-negative patients. No H. pylori strains were resistant to AMX or TET in either group. Resistance to CLA was more frequent in patients with a CD4 count < 200 cells/μL than in those with a CD4 count ! 200 cells/μL (3/7 (42.8%) vs. 10/84 (11.9%), p = 0.05).
Black Africans who were HIV-positive had a significantly greater incidence of resistant strains than did Black Africans who were HIV-negative (p = 0.04, Table 3). Within the other ethnicities, HIV-positive patients were more likely to have resistant strains than HIV-negative patients, but the differences were not statistically significant. HIV-positive women were more likely to have resistant strains than HIV-negative women (p = 0.001), notably for LEV (p = 0.001) and MTZ (p = 0.03). There were no significant differences in the incidences of resistance in male HIV-positive and -negative patients.
In a sub-analysis, the prevalence of H. pylori resistant strains was assessed for each antibiotic according to HIV status and then stratified according to ethnic group. These analyses showed that resistant strains were significantly more common among Black African patients than among North African and Caucasian Europeans  In multivariable analysis in the whole cohort, ethnicity and HIV status were independent predictors of H. pylori resistance (Table 4). In the subgroup of HIV-infected patients, sex and AIDS were independent predictors of H. pylori resistance (Table 5)

Discussion
To our knowledge, this is the first prospective study in which the susceptibility of H. pylori strains to antibiotics has been compared in HIV-positive and HIV-negative patients [14].
Overall, resistance to MTZ was most frequent, followed by LEV and CLA resistance. Our results are similar to those of a previous multicenter study carried out among unselected patients in Belgium, but differ from studies conducted in Africa, where a much higher prevalence of resistant strains has been found [17,20,21]. Our major finding was the significantly greater prevalence of primary H. pylori resistance to MTZ and LEV in HIV-positive patients compared to HIV-negative patients, and a trend towards a greater prevalence of CLA resistance.
In multivariable analysis, HIV status was a strong and independent predictive factor for resistance of H. pylori. It is possible that the higher prevalence of H. pylori resistance in HIVpositive patients could be due to greater previous cumulative antibiotic exposure, either for treatment of acute infections or for chemoprevention of opportunistic diseases. However, except for the patients with AIDS, HIV-positive patients in this study had less frequently received antibiotics in the 12 months prior to H. pylori diagnosis than had the HIV-negative patients. It has been suggested that twice as many Clostridium difficile infections (often treated with MTZ) occur in HIV-positive patients (notably with AIDS or lower T CD4 counts or postantibiotic exposure) than in HIV-negative patients [22,23]. However, we were unable to determine whether this was the case in these patients. HIV-positive patients also had a higher prevalence of H. pylori resistant to LEV than HIV-negative patients; this has not been documented previously. Fluoroquinolone use may be higher in HIV-positive than in HIV-negative patients because of its use as empiric treatment for infectious diarrhea, urinary tract infection, or pneumonia (25-fold), which are more common in HIV-positive patients [24,25]. Finally, the prevalence of H. pylori CLA resistance was nearly 14% in HIV-positive patients compared to 10% in HIV-negative patients. The frequency of H. pylori CLA resistance in HIV-positive patients was higher than that recently found among unselected patients in Belgium [17]. This greater prevalence could be, at least in part, due to the use of macrolides in the chemoprevention of Mycobacterium avium complex (MAC) infection in severely immunocompromised HIV patients (CD4 counts <100). Interestingly, we found that CLA resistance was more frequent in HIVpositive patients with CD4 counts <200 cells/μL than in those with higher CD4 counts. Use of macrolides for other indications, including respiratory infections with atypical microorganisms and sexually transmitted diseases such as chlamydiosis, may also be involved [1,26]. HIV-positive patients had more multidrug-resistant H. pylori strains than did HIV-negative patients.
Interestingly, AIDS was an independent predictive factor for carrying H. pylori resistant strains. This finding has not been published previously. The explanation could be that, as we have demonstrated, patients with AIDS were exposed to more antibiotics, which could potentially lead to H. pylori resistance. Moreover, in our hospital, 20% of patients newly diagnosed with HIV infection have a CD4+ count < 200 and are likely to receive antibiotics to cure or prevent infection (S. De Wit, personal communication).
Another independent risk factor for H. pylori resistance was ethnicity. Indeed, we demonstrated that the prevalence of resistant strains was significantly different across ethnicities. Resistant H. pylori strains were more frequently isolated from patients of Black African ethnicity than from those of North African or Caucasian European ethnicities. This was particularly true for MTZ, LEV, and multidrug resistance in Black African patients who were HIV-positive. The frequent use of imidazole derivatives for the treatment of gynecological or protozoal infections in Africa, the region from which these patients originated, may play a role in this increased MTZ resistance. Alternatively, a subset of Black African patients may have been infected with H. pylori in Africa (an endemic region for H. pylori) with a strain that was MTZ resistant without previous exposure to this drug [20,21]. However, this explanation is less likely because the prevalence of MTZ resistance among Black Africans in our study was less than that found in some African regions in which virtually all patients are infected with MTZresistant strains [15,20,21]. The prevalence of H. pylori strains resistant to LEV was higher among HIV-infected Black Africans than among HIV-infected patients from other ethnicities. It is possible that Black African HIV-positive patients may have been exposed more commonly to LEV for appropriate indications, such as those already mentioned above (urinary or digestive infections) because they had more pronounced immune depression (lower CD4 counts) and a higher risk of pneumonia, or they may have been exposed to inappropriate prescription of fluoroquinolones [15,20,21,24,25].
Overall, multidrug-resistant strains were more common among patients of Black African ethnicity than North African and Caucasian European ethnicities. This was particularly true among HIV-positive black African patients [17,20,21]. HIV-positive Black Africans may have had more cumulative antibiotic exposures than other ethnicities given that they had significantly lower median CD4 counts.
Sex was also an independent risk factor for H. pylori resistance among HIV-positive patients in contrast to the whole cohort. HIV-infected females were more commonly infected with H. pylori strains resistant to LEV and MTZ. Female sex is associated with more frequent urinary infections and gynecologic infections, which are often treated with LEV and MTZ, respectively. This is an additional risk factor that could be exacerbated in the presence of HIV/AIDS. Sex has been reported as a risk factor for resistance in other studies in the general population [14,17,27].
Another interesting finding is the absence of H. pylori with resistance to AMX or TET in all patients enrolled. The absence of H. pylori resistance to AMX or TET was also reported in a multicenter study of unselected patients from Belgium [17]. In contrast, H. pylori strains resistant to AMX and TET have been described in Africa [15,20,21].
Overall, the most commonly prescribed treatment regimen was triple therapy. HIV-positive and HIV-negative patients in our study achieved lower eradication rates compared to previous reports, despite good tolerance [13]. The end-point of this study was not the effectiveness of anti-H. pylori treatment, so we cannot draw any definitive conclusions on this aspect from our results. Nevertheless, we observed a lower response rate among HIV-positive patients treated with antiretroviral drugs compared to HIV-positive patients who were not treated with antiretrovirals; it may, therefore, be interesting to evaluate the role of higher CD4 count (a risk-factor for H. pylori infection) on response rates as well as the possibility of an interaction between anti-H. pylori and antiretroviral drugs on treatment response [2,28]. Several arguments have emerged from the current literature that may explain, in part, the suboptimal response to anti-H. pylori treatment seen in our patients [8][9][10][11][12][13][29][30][31][32]. Antibiotic-resistance is the major cause of failure to eradicate H. pylori. CLA-resistant strains are, for example, associated with a 70% reduction in efficacy of H. pylori triple therapy. There was a high prevalence of H. pylori CLAresistant strains among HIV-positive patients, nearly reaching the threshold of 15 to 20% at which the use of this drug in the treatment against H. pylori should be avoided [8]. There was also variation in the duration of triple therapy in this study, ranging from 7 to 14 days, and duration of treatment has been shown to impact on the effectiveness of triple therapy against H. pylori [8][9][10][11][12][13].
The strengths of this study include the prospective collection of data, and the enrollment of a large number of patients who were all naïve to H. pylori treatment and had antibiotic susceptibility results available. Moreover, patients were well-matched for age and were from different ethnicities. The study provides insight into the chemosusceptibility of H. pylori strains among subpopulations, notably those who were HIV-positive. Limitations of this study include the fact that the HIV-positive North African group was underrepresented. However, this pattern reflects the current epidemiological trend in our hospital. In addition, treatment for H. pylori infection was not studied extensively because effectiveness of treatment was not the end-point.

Conclusion
The data presented in this paper demonstrate that there is a higher prevalence of primary H. pylori resistance in HIV-positive compared to HIV-negative patients. In these patients, primary resistance was most frequently to CLA, LEV, and MTZ. Multidrug resistance was also frequently found, notably among Black Africans. Among HIV-infected patients, sex and a diagnosis of AIDS were independent predictors of H. pylori resistance.
A careful history of antimicrobial use should be taken and results of susceptibility testing awaited in HIV-infected patients and individuals from developing countries before anti-H. pylori therapy is started. If antimicrobial susceptibility testing is not available, we suggest taking into account the patient's risk profile for resistance and avoiding regimens that include clarithromycin or fluoroquinolones. Overall, these data provide evidence that the assessment of H. pylori drug resistance is warranted in HIV-positive patients and further studies in other populations/regions are desirable.