Diagnosis and Epidemiology of Helicobacter pylori Infection
Abstract
A limited amount of new information was published in the field of diagnosis and epidemiology of Helicobacter pylori this last year. Besides some improvement in current tests, it is interesting to note the attempts to identify severe disease, for example gastric cancer, by breath analysis using nanomaterial-based sensors. In contrast, the predictive value for gastric cancer and atrophy of pepsinogen determinations was found inadequate.
Prevalence studies of H. pylori infection have been carried out in adults and children around the world in the general population but also in specific communities. The usual risk factors were found. In addition, a Japanese study highlighted the role of grandmothers in the familial transmission of H. pylori.
A study showed that the infection may not always readily establish itself in children, given the number of transient infections observed. It was also noted that after eradication, a first-year relapse is likely to be a recurrence of the previous infection, while later on it is probably a reinfection with a new strain.
Diagnosis
General Evaluation of Helicobacter pylori Diagnostic Tests
Two studies evaluating different diagnostic methods in specific settings concluded that histology was the most accurate test: Choi et al. 1 evaluated several tests for detecting H. pylori in Korean patients with bleeding (n = 157) and nonbleeding (n = 247) peptic ulcer. In bleeding patients, the sensitivity and specificity were 92.5 and 96% for histology, 40 and 100% for culture, 85 and 92% for rapid urease test (RUT), and 97 and 56% for serology, respectively. The sensitivities of culture and RUT in bleeding patients were significantly lower than in nonbleeding patients. By contrast, histology was reliable regardless of bleeding. Tian et al. 2 reported a meta-analysis evaluating H. pylori diagnostic methods in patients with partial gastrectomy. The pooled sensitivity and specificity were 93 and 85% for histology, 77 and 89% for urea breath test (UBT), and 79 and 94% for RUT, respectively. They concluded that histology was the most reliable test in this setting.
Invasive Tests
Endoscopy
A multicenter Japanese study evaluated the possibility to diagnose H. pylori by conventional endoscopy and chromoendoscopy using indigo carmine in comparison with histology performed according to the Sydney System, on 275 patients 3.
Based on a set of indices including diffuse and spotted redness, enlarged and tortuous folds, and swelling, they achieved a sensitivity of 94% in the corpus and 88% in the antrum. However, the specificity was only 62 and 52%, respectively.
Rapid urease test
Moon et al. 4 studied 240 patients to compare the efficacy of RUT performed on separate gastric antrum and corpus tissue specimens versus on combined specimens, using histology as a gold standard. Combining tissues prior to RUT increased the detection of H. pylori from 137 (64%) in separate specimens to 148 (69.2%) in combined specimens (p < .01). Furthermore, positive reactions in combined specimens required less time.
Histology
Lee et al. 5 evaluated 91 patients requiring endoscopic mucosal resection for early gastric cancer (GC). They obtained three pairs of biopsies from the antrum, corpus lesser curvature (CLC), and corpus greater curvature (CGC). The sensitivity of histology in detecting H. pylori was significantly higher in the CGC than that in the antrum or CLC (84.8 vs 30.3 and 47.0%, respectively; p < .001). They concluded that the CGC is the optimal biopsy site for H. pylori in patients with extensive atrophy.
PCR
PCR was among the diagnostic techniques most frequently evaluated in 2012. An interesting study came from Portugal: Ferreira et al. 6 reported a novel method for genotyping the vacA intermediate (i) gene region. They proposed a novel primer combination allowing the amplification of smaller DNA fragments than the original PCR, and therefore, it can be applied to paraffin-embedded biopsies. It was effective in characterizing the vacA i region in 191 (99.5%) of 192 frozen and in 186 (95.9%) of 194 formalin-fixed paraffin-embedded gastric biopsies. The vacA i1 strains showed an increased risk for gastric carcinoma (OR, 22; 95% CI, 7.9–63). This method may be useful for detecting patients at high risk of cancer who would be suitable for H. pylori treatment and/or close follow-up.
Ou et al. 7 evaluated the accuracy of a new fluorescence quantitative PCR (fqPCR) for H. pylori detection in 138 children. This technique detected infection in all 38 patients also positive with conventional tests and in 8 of the 100 patients with negative conventional tests. They concluded that fqPCR was more sensitive than histology, RUT, or UBT. Similarly, one Spanish group published two reports 8, 9 evaluating quantitative real-time PCR in 158 consecutive patients with digestive disorders, 80 of whom had gastrointestinal bleeding. They found high sensitivity (95%) but low specificity (73%) for PCR on antral biopsies from bleeding patients. These authors suggest that many of the “false-positive” PCR results may correspond to true infections undetected by conventional tests. Monno et al. 10 compared the detection of clarithromycin-resistant H. pylori by a TaqMan real-time PCR (RT-PCR) performed on paraffin-embedded gastric biopsy samples to the results of E-test performed on strains isolated from the corresponding patients.
Ten of the 41 cases detected as susceptible by E-test were resistant by RT-PCR, of which eight presented a mixed population of susceptible and resistant strains. However, surprisingly, the TaqMan RT-PCR could not detect 7 of the 47 clarithromycin-resistant strains according to E-test, indicating that a technical improvement must be done on this PCR format.
Noninvasive Tests
Urea breath test
Olafsson et al. 11 evaluated 620 UBT in 595 subjects at a clinical gastroenterology practice. They collected detailed information about recent treatment with proton-pump inhibitors (PPI), antibiotics, or bismuth consumption. They also collected data about H. pylori eradication treatment finished <4 weeks before UBT and previous gastric resection. UBT was negative in 526 patients. However, 120 (23%) of the negative results occurred in patients receiving one or more of these previous treatments. Moreover, in approximately 45% of those tested after eradication treatment, UBT was performed <4 weeks before the end of the therapy. The authors suggest that protocol violations are frequent in clinical practice, leading to potential decreases in the reliability of UBT.
Velayos et al. 12 investigated the accuracy of UBT performed immediately after emergency endoscopy in patients with peptic ulcer bleeding (PUB) by comparing the results with those of UBT performed after hospital discharge, considered the gold standard. The sensitivity and specificity of the early UBT were 86 and 66%, respectively, with a negative predictive value of 50%. The authors conclude that the low negative predictive value of the early test makes a delayed test mandatory. This conclusion is in agreement with the PUB consensus recommendations 13.
Stool tests
Sato et al. 14 evaluated two new stool tests in 111 fecal samples: an enzyme-linked immunosorbent assay (ELISA) (Testmate pylori antigen enzyme immunoassay TPAg EIA) and an immunochromatographic assay (Testmate Rapid Pylori Antigen (Rapid TPAg)) using a monoclonal antigen against H. pylori catalase. They reported an outstanding accuracy of both Testmate kits of 100%.
Noninvasive Diagnosis of Gastric Cancer
The quest for noninvasive methods to diagnose GC, the most feared complication of H. pylori infection, is one of the most active areas of research. Cao et al. 15 determined serum anti-H. pylori immunoglobulin G (IgG), pepsinogen I (PGI), and pepsinogen II (PGII) by ELISA in 450 patients with GC, 111 individuals with gastric atrophy, and 961 healthy controls. PGII levels were higher in patients with GC or gastric atrophy than in controls. For atrophic gastritis, the best cutoff value was 82.30 μg/L (sensitivity 85.9%, specificity 75.1%), and for PGI/PGII ratio, the best cutoff value was 6.05 (sensitivity 78.3%, specificity 71.6%).
Shikata et al. 16 prospectively followed 2446 Japanese men and women aged ≥40 years for 10 years, 69 developed GC during follow-up. The best cutoffs to discriminate GC were PGI ≤59 ng/mL and PGI/II ratio ≤3.9. The sensitivity and specificity for predicting GC were 71.0 and 69.2%, respectively.
As the poor accuracy of pepsinogen determinations for either atrophy or GC probably makes this measurement inadequate for screening, new methods are in development. Two of these are especially interesting:
Li et al. 17 investigated three microRNAs (miR-223, miR-21, and miR-218) in 60 patients with GC and 60 healthy controls. Combined ROC analysis yielded an excellent area under the curve value (0.9531) for discriminating patients with GC from healthy controls. The miRNAs were also elevated in patients with early GC. These promising results require further validation.
Finally, the most striking study this year came from an Israeli–Chinese collaboration; Xu et al. 18 evaluated exhaled breath samples from 130 patients (37 with GC, 32 with ulcer, and 61 with less severe conditions) who underwent endoscopy and biopsy. Breath samples were analyzed using nanomaterial-based sensors. Predictive models based on breath analysis achieved excellent discrimination between GC and benign gastric conditions (89% sensitivity, 90% specificity). Chemical analysis found that five volatile organic compounds (2-propenenitrile, 2-butoxyethanol, furfural, 6-methyl-5-hepten-2-one, and isoprene) were significantly elevated in GC and/or peptic ulcer compared with less severe gastric conditions.
Epidemiology
Prevalence of Helicobacter pylori Infection
Adults
Two large studies were performed in Eastern Europe (Table 1). The Czech study aimed to determine the prevalence of H. pylori infection and analyzed a sample representative of the whole country. An important decrease in prevalence from 2001 to 2011 was noted 19. The Latvian study focused on the determination of atrophy using the PGI/II ratio and analyzed sera from older subjects. The prevalence was as high as 79.2% in line with the high GC rate in this country 20. A high prevalence was also found in the six Latin America countries explored, that is, Chile, Columbia, Costa Rica, Honduras, Mexico, and Nicaragua (75–83%) without a decrease in younger age cohorts 21.
| Country | Years specimen collection | Number | Method | Prevalence % | Authors |
|---|---|---|---|---|---|
| (1) General population | |||||
| Czech Republic | 2011 | 1441 | UBT | 23.5 | Bures J et al. 19 |
| Latvia | 2008–2009 | 3564 | Serology | 79.2 | Leja M et al. 20 |
| Latin Americaa | 2009–2012 | 1852 | UBT | 79.4 | Porras et al. 21 |
| USA (Texas) | 2011 | 281 | Serology | 17.1 | Patterson et al. 50 |
| Japan (Nagoya) | 2008–2010 | 5167 | Urine Ab | 39.6 | Tamura et al. 22 |
| Saudi Arabia (Al Madinah) | 456 | Serology | 28.3 | Hanafi et al. 23 | |
| Morocco (Fez)b | 2009 | 429 | Biopsy based tests | 75.5 | Benajah et al. 47 |
| (2) Specific groups | |||||
| Korea (Vietnamese female migrants) | 2006–2007 | 390 | Serology | 55.7 | Baik et al. 24 |
| China (North, migrants) | 2007–2011 | 324 | RUT | 41.5 | Xia et al. 25 |
| Nepal (Sherpas) | 383 | SAT | 70.5 | Sherpa et al. 26 | |
| Mexico (Mennonites) | 152 | Serology | 50.7 | Alvarado-Esquivel et al. 28 | |
| Mexico (Tepehuanos) | 2010–2011 | 156 | Serology | 66 | Alvarado-Esquivel et al. 27 |
| USA (Obese) | 2001–2009 | 611 | Histology | 23.7 | Verma et al. 30 |
| Canada (First Nations population)b | 2009–2011 | 203 | Histology | 37.9 | Sethi et al. 29 |
- UBT, urea breath test; SAT, stool antigen test; RUT, rapid urease test; Ab, antibody;
- a includes Chile, Columbia, Costa Rica, Honduras, Mexico, and Nicaragua.
- b Patients.
Among the risk factors explored in all these studies, the results always pointed out the same, that is, low socioeconomic status including limited education, crowded homes, and difficult access to sanitized water. The prevalence by age decreases in relation to the socioeconomic development as is seen in Japan 22, the Czech Republic 19, and Saudi Arabia 23. No gender difference was observed, but surprisingly in the Czech Republic, the status of being married or divorced was a risk factor. Several studies were carried out in special populations. Female immigrants from Vietnam are common in Korea. The prevalence of H. pylori infection in these Vietnamese women was indeed lower (55.7%) than that of the Korean females (71.4%) 24. Migrant workers in large cities of Northern China were also tested for H. pylori infection and had a surprisingly low rate of infection (41.5%) 25.
In Eastern Nepal, an isolated Sherpa community had a prevalence of 70% 26. Two specific groups from Mexico were also explored, indigenous people living in remote rural communities, the Tepehuanos 27, and a group of German descent, also living in rural communities as farmers, the Mennonites 28. The global prevalence observed within these two groups was 66 and 50.7%, respectively. First Nations populations in Northwestern Ontario had a lower prevalence than expected (37.9%) even if, in contrast to the other studies, symptomatic subjects were explored 29. However, the lowest prevalence was observed among the obeses in the USA (23.7%) 30.
A study carried out on 665 persons from the Southern Community cohort recruited in 12 southeastern states of the US, addressed the interesting question of the role of neighborhood measures of socioeconomic status (based on geolocalization) to assess differences in H. pylori prevalence, especially CagA+ H. pylori, between individuals of African descent (89%) and Caucasians (69%), both of low socioeconomic status. The neighborhood-level measures of education, employment, and home values were associated with CagA+ H. pylori prevalence but did not explain the persistent and strong relationship found with African descent, leading to the hypothesis that the link may have a biological basis 31.
Children
Four studies were performed in Western Europe (Table 2). In contrast to the data from the Czech Republic 19 and Italy 32, 33, those from Portugal 34 pointed to a very high prevalence of H. pylori infection in 13-year-old children, implying that H. pylori-associated diseases may still be present for a while in this country. This rate was even the highest of all of the studies, ranking above both Iran 35 and the Bahamas in the Caribbean 36, despite the latter high standard of living.
| Country (city) | Recruitment | Years specimen collection | Number | Method | Prevalence % | Authors |
|---|---|---|---|---|---|---|
| Czech Republic | Population-based (5–14 year) | 2011 | 385 | UBT | 4.9 | Bures J et al. 19 |
| Italy (Sardinia) | School children (6–15 year) | 1996–1998 | 1741 | Serology | 13.3 | Dore et al. 32 |
| Italy (Rome) | School children (3–10 year) | 400 | SAT | 8.7 | Sabbi et al. 33 | |
| Portugal (Porto) | School children (13 year) | 2003 | 1312 | Serology | 66.2 | Bastos et al. 34 |
| Bahamas (Nassau) | School children (6–12 year) | 2008 | 96 | UBT | 54.2 | Carter et al. 36 |
| Mexico (Mexico City) | School children (6–13 year) | 2005–2010 | 940 | UBT | 38 | Duque et al. 39 |
| Nigeria (Uyo) | Emergency Unit (6 m–15 year) | 2008 | 230 | Serology | 30.9 | Etukudo et al. 38 |
| Uganda (Kampala) | Population-based (0–12 year) | 2007 | 427 | Serology | 44.3 | Ankarklev et al. 37 |
| Iran (Kurdistan) | Population-based (4 m–15 year) | 2007 | 458 | SAT | 64.2 | Jafar et al. 35 |
| Pakistan (Karachi) | Population-based (1–15 year) | 540 | Serology | 47.2 | Jafri et al. 51 |
- UBT, urea breath test; SAT, stool antigen test.
There was no protection from breastfeeding 35, 36. In one study, an association was found with Giardia intestinalis suggesting a common transmission route 37. When the prevalence was presented by year, a transient decrease was observed instead of a constant increase 35, 38.
Duque et al. 39 followed a large number of children with UBT (718) for 6 months after the first breath collection. A change in H. pylori status was observed in 84 children (11.7%). Of the 61 new cases, 24 were transient, 18 persistent, and 19 undetermined, and of 23 spontaneous clearance cases, six were transient, seven persistent, and 10 undetermined.
These data as well as others indicate that H. pylori may not readily establish a chronic infection in some cases and that the host children may spontaneously eliminate the bacterium after acute infection. This area deserves more investigation.
Transmission
The routes of transmission of H. pylori still remain unclear. Person-to-person transmission and intrafamilial spread seem to be the main route, based on the intra-familial clustering observed, while a waterborne infection remains possible.
Person-to-person transmission
Urita et al. 40 studied person-to-person H. pylori transmission in a small town in Japan where the family structure is still traditional. Of the 838 children tested, 101 were H. pylori positive (12.1%). They determined the H. pylori status of the family members of these index cases, and a logistic regression showed that siblings carried the main risk, followed by mothers and grandmothers but not by fathers and grandfathers. The role of grandmothers who may take care of their grandchildren when mothers are at work was highlighted for the first time.
The possibility of H. pylori transmission among children in day-care centers or kindergarten where interpersonal contacts are common was also proposed. A meta-analysis of 16 studies did not confirm this hypothesis. The summary OR was 1.12 [95% CI: 0.82–1.52]. However, the authors highlighted the limitations of the published studies including different types of childcare, different age groups, and lack of differences in the exposure duration, with exposure giving a high heterogeneity to the meta-analysis results 41.
Environmental transmission
Only one article previously reported H. pylori isolation in water 42. Moreno and Ferrus were able to culture H. pylori from 6 of 45 wastewater samples by applying a modified filtration method to eliminate competitive microbiota; 21 other samples that could not be cultured were positive by PCR or FISH 43.
Another study claimed the isolation of H. pylori from five treated water samples originally from a river in Isfahan, Iran (two from tap water, two from a dental unit's water, and one from a public water cooler, but none from bottled water). In addition, 14 samples were positive by PCR 44.
In the metropolitan area of Karachi, Pakistan, 2 of 50 drinking water samples tested were found to be positive for H. pylori by PCR 45.
Reinfection after Treatment
After successful eradication, usually confirmed by a UBT, at least 4 weeks after the end of the treatment, a relapse of the infection can occur. Take et al. 46 tried to determine whether such a relapse, which occurred in 26 of 1609 patients during a mean follow-up of 4.7 years, was a recurrence of the previous infection or a reinfection by a new strain. They compared the strains by random amplification of polymorphic DNA fingerprinting and showed that during the first year after the successful treatment, 6 of 10 strain couples were genetically identical, while none of the four strain couples tested at a later follow-up were identical. In 12 cases, the comparison could not be performed. This study confirmed that first-year relapses are more likely to be a recurrence of the same infection.
Nevertheless, such a recurrence may occur at various rates. In Morocco, only two patients relapsed (0.8%): 1 of 239 at 6 months and 1 of 222 after a year 47, while in Latin America, 1-year relapse rate was 125 of 1091 (11.5%) patients. The risk factors identified were noncompliance to the initial therapy, presence of children in the household, and also the study center 48.
The most fascinating study of this year concerns the ability of molecular fingerprinting (MLST) to follow the spread of H. pylori by our ancestors from Africa. Moodley et al. 49 estimated that H. pylori is approximately as old as modern humans and that migration out of Africa occurred in several waves, the first one 60,000 years ago and the second 52,000 years ago.
Acknowledgements and Disclosures
Competing interests: the authors have no competing interests.
