Virulence-Associated Genotypes of Helicobacterpylori: Do They Explain the African Enigma?Luis E. Bravo, M.D., Leen-Jan van Doorn, Ph.D., J. Luis Realpe, M.D., and Pelayo Correa, M.D.Department of Pathology, Universidad del Valle, Cali, Columbia; Delft Diagnostic Laboratory, Delft, TheNetherlands; Department of Gastroenterology, Hospital Departamental, Pasto, Colombia; and Department ofPathology, Louisiana State University, New Orleans, Louisiana

OBJECTIVES: The aim of this study was to compare thedistribution of virulence-associated genotypes of Helicobac-ter pylori in two Colombian populations with contrastinggastric cancer risk but with similar H. pylori infection prev-alence.

METHODS: Gastric biopsies were taken from 241 subjectsfrom the high gastric cancer risk area of Pasto and from 93subjects from the low risk area of Tumaco. Four gastricbiopsies from each patient were fixed in 10% bufferedformalin for histopathologic analysis, and one was frozenimmediately in liquid nitrogen and used for genotyping.CagA and vacA genotypes were determined by multiplexpolymerase chain reaction and reverse hybridization on aline probe assay.

RESULTS: In patients from the population with high risk forgastric cancer, statistically significant higher relative fre-quencies of cagA positive and vacA s1 and m1 genotypeswere found as compared to the population from the low riskarea.

CONCLUSIONS: Although H. pylori infection has been rec-ognized as a cause of gastric cancer in humans, some largepopulations with high prevalence of infection have lowgastric cancer rates. This so-called “African enigma” so farremains unexplained. Our findings suggest that virulence-associated genes of H. pylori may partially explain theAfrican enigma. Other factors, including human geneticpolymorphisms and diet, are also suspected to play a majorrole. Further investigations are needed to test thishypothesis. (Am J Gastroenterol 2002;97:2839–2842.© 2002 by Am. Coll. of Gastroenterology)

INTRODUCTION

Although Helicobacter pylori infects approximately half ofthe world’s population, only a small proportion of infectedsubjects develop symptoms or clinically significant disease.The so-called “African enigma” refers to the fact that theprevalence of H. pylori colonization in that continent ishigh, but the frequency of gastric pathology, especiallygastric cancer, is very low (1). The reasons for this fact areunknown but may be related to the host’s immunological

defenses, environmental factors such as food, and/or to thevirulence of different strains of the bacteria. Several viru-lence-associated genes of H. pylori have been identified.CagA is a marker for a genomic pathogenicity island ofabout 40 kilobase pairs, whose presence is associated witha more severe clinical outcome (2). CagA positive strainsinduce more severe inflammatory damage in the gastricmucosa as compared to cagA negative strains. The CagAprotein by itself does not induce inflammation in experi-mental animals, and it is not clear which genes encoded bythe pathogenicity island are responsible for the inflamma-tion (3). CagA positive strains are associated with highergastric cancer risk and with duodenal ulcer, although theseoutcomes of the infection seem to be mutually exclusive (4,5). The CagA protein itself is translocated into epithelialhost cells by a type IV secretory apparatus (6, 7).

The vacuolating cytotoxin vacA gene is also associatedwith virulence. VacA is polymorphic in its s and m regions(s and m), and these regions display marked and distinctvariants (8). The mosaic composition of the vacA s and mregions is associated with different disease outcomes (8).Moreover, vacA subtypes have a particular geographic dis-tribution (9). Subtype s1a is predominant in populations ofnorthern European ancestry, and it has been suggested thatit is associated with duodenal ulcer (9, 10). Its associationwith gastric cancer is unknown. Subtype s1b is predominantin Africa and is also very frequently found in Portugal,Spain, and Central and South America. In France, Italy, andthe United States, the frequency of s1a and s1b genotypes issimilar. VacA m genotypes are also polymorphic, m1 beingpredominant in most populations studied. The different mtypes seem to recognize different receptors on epithelialcells and may induce different intracellular responses.

In Colombia, the contrasts between infection and diseaseare clear and marked. The prevalence of H. pylori infectionis very high throughout the country. In general, populationsliving in the mountains have high rates of gastric cancer,whereas those from the Atlantic and Pacific coasts have verylow rates (11, 12).

The population-based Cancer Registry from Cali, Colom-bia reported that migrants from Narino had an age-adjustedincidence rate of gastric cancer during 1962–66 of 116 per

THE AMERICAN JOURNAL OF GASTROENTEROLOGY Vol. 97, No. 11, 2002© 2002 by Am. Coll. of Gastroenterology ISSN 0002-9270/02/$22.00Published by Elsevier Science Inc. PII S0002-9270(02)05448-5

100,000 inhabitants, compared to 41.9 for Cali natives (13).Narino is the southernmost “department” (state) of Colom-bia and has a high altitude mountain area (�2800 m abovesea level), where the towns of Pasto and Tuquerres arelocated, and a low altitude region whose main city is Tu-maco, on the Pacific coast. A case-control study of gastriccancer done between 1968 and 1972 reported a relative riskof gastric cancer of 2.1 for subjects born 2000 m or higherabove sea level, compared to 0.47 for subjects born between1000 and 2000 m above sea level. An estimate of gastriccancer incidence for 1972–1974 for the high altitude moun-tains of Narino reported the highest known rate: 150 per100,000, adjusted to the world population (14). The inci-dence rate of gastric cancer in Cartagena, on the Atlanticcoast, in 1966 was 2.8 per 100,000 for men and 2.4 forwomen (15).

An autopsy study done in Cali from 1962 to 1966 regis-tered a prevalence of metaplasia of 58.4% in immigrantsfrom Narino, compared to 21.3% for immigrants from theAtlantic and Pacific coasts (11). A similar study in Carta-gena reported a prevalence of metaplasia of 11% (13).

The prevalence of H. pylori infection in adults from thehigh mountains of Narino was reported as more than 90%(16). In Cartagena, a study carried out in the UniversityHospital reported a prevalence of H. pylori infection of 50%in symptom-free subjects, 98% in duodenal ulcer patients,and 90% in gastric ulcer patients (17).

MATERIALS AND METHODS

We examined the distribution of vacA and cagA genotypesof H. pylori in gastric biopsies from two populations withcontrasting cancer risk. The first group comprised individ-uals living at high altitude in the Andes Mountains in thetowns of Pasto and Tuquerres, where rates of gastric cancerare very high. They are predominantly mestizos of Spanish-Amerindian ancestry and are mostly agricultural rural work-ers. The low risk group comprised individuals from thePacific coast town of Tumaco, where the incidence of gas-tric cancer is low (14). They are predominantly of African-American ancestry, and their occupations are mostly relatedto fishing.

Gastric biopsies were taken from adult volunteers. Gas-troscopy was performed at Hospital Departamental in Pastoand at Hospital San Andres in Tumaco. Four gastric biopsieswere immediately fixed in 10% buffered formalin for his-tological examination: one from the lesser curvature of theantrum, one from the greater curvature of the antrum, onefrom the incisura angularis, and one from the anterior wallof the corpus. One additional biopsy taken from the greatercurvature of the antrum was immediately frozen in liquidnitrogen and transported in dry ice to Delft DiagnosticLaboratory in the Netherlands. Genotypes of H. pylori col-onizing the gastric mucosa were determined by multiplexpolymerase chain reaction and reverse hybridization onto aline probe assay as previously described (9, 18–22). The

ethics committee of both hospitals and the ethics committeeof Universidad del Valle, Cali, Colombia, approved thisstudy. Informed consent was given by each participant.

RESULTS

A total of 334 patients from two geographic origins inColombia were studied. The high risk group comprised 241patients from the Pasto region. The low risk group com-prised 93 patients from the Tumaco region.

The prevalence of H. pylori as determined by a modifiedWarthin-Starry Sliver stain done in all gastric biopsies wasvery similar in both populations: 73.86% in Pasto versus79.56% in Tumaco. Table 1 shows the histopathologic di-agnoses in infected subjects. A higher prevalence of non-atrophic gastritis in Tumaco and of intestinal metaplasia anddysplasia in Pasto can be seen. The relative frequency ofmultifocal atrophic gastritis and intestinal metaplasia wassignificantly higher in Pasto than Tumaco: OR 2.2 (95% CI1.59–2.97).

The distribution of vacA and cagA genotypes in biopsieswith adequate material for genotyping is shown in Table 2.In both populations the vacA s1b genotype accounts for over95% of all s1 strains. The prevalence of cagA positivestrains (excluding subjects with infection by multiplestrains) was significantly higher in patients from the Pastoregion (90.4%), as compared to those from the Tumacoregion (81.1%) (�2 test, p � 0.043). Similarly, the preva-lence of vacA s1 strains was significantly higher in patientsfrom the Pasto region (93.2%), as compared to those fromthe Tumaco region (83.7%) (�2 test, p � 0.011). CagApositivity and vacA s1 genotypes were strongly correlated inboth populations (�2 test, p � 0.001). The prevalence ofvacA m1 genotypes was higher in Pasto (83.3%) than inTumaco (70.2%) (�2 test, p � 0.013). All three markers ofvirulence, namely cagA, vacA s1 and m1, are significantlyhigher in frequency in the high risk (Pasto) than in the lowrisk (Tumaco) area.

DISCUSSION

We investigated H. pylori positive patients from two geo-graphic areas in Colombia. The great majority of patientsharbored strains of the vacA s1b subtype. This is in agree-

Table 1. Histopathologic Diagnosis in H. pylori PositiveSubjects from Pasto and Tumaco Regions of Colombia

Diagnosis Pasto Tumaco

NAG 16 (9.0) 40 (54.05)MAG 30 (16.9) 13 (17.57)MAG-IM 100 (56.2) 14 (18.92)Dysplasia 31 (17.4) 0Gastritis–NOS 1 (0.6) 7 (9.46)Total 178 74

Data are expressed as n (%).IM � intestinal metaplasia; MAG � multifocal atrophic gastritis; NAG � non-

atrophic gastritis; NOS � undermined (insufficient material).

2840 Bravo et al. AJG – Vol. 97, No. 11, 2002

ment with earlier observations that most H. pylori strainsfrom South America contain the s1b genotype, whereassubtype s1a is rare (9). Because this subtype is predominantin high as well as in low gastric cancer risk countries, it isnot surprising that it is also predominant in high as well asin low risk areas of Southern Colombia; s2 genotypes areless common, and their relation to gastric pathology anddisease outcome is unknown.

In our sample, m1 genotypes were more prevalent inPasto, where gastric cancer incidence is high, whereas cagAnegative, vacA m2 genotypes were relatively more frequentin Tumaco, where gastric cancer rates are very low. Becausethe cagA positive, vacA s1 and m1 genotypes are stronglyassociated, it is difficult to assess the relative importance ofeach genotype marker.

Our results indicate that genotypes with virulence deter-minants are more frequent in a population with a high riskfor gastric cancer, as compared to a population with a lowrisk. This observation would offer a partial explanation forthe African enigma: a high prevalence of H. pylori infectionbut a low incidence of gastric cancer. The differences ingastric cancer rates, however, cannot be completely ex-plained by the difference in the higher prevalence of lessvirulent strains, indicating that additional etiological factorsalso play a role. There are major differences in the racial andcultural backgrounds of the residents of the high Andesmountains (predominantly Amerindian) and the PacificCoast of Colombia (predominantly African-American). Thisracial difference, however, may not explain the differencebetween populations of native Africans (low cancer risk)

and migrant Africans to America (higher risk). Polymor-phisms in host cytokines, such as interleukin 1� (23), mayalso be an important risk factor for H. pylori-associatedgastric carcinoma.

We recently studied genotypes in infected patients inPasto in whom triple therapy failed to eradicate the infection(24). The study showed that in strains surviving such treat-ment, those with low virulence markers, namely cagA neg-ative and vacA s2/m2 were relatively more frequent. Morevirulent strains appeared to be more susceptible to antimi-crobials. There are major differences between the two Co-lombian populations studied here with respect to diet. Thehigh altitude Andes residents’ diet is very rich in grains(mostly corn) and roots (mostly potatoes). The coastal res-idents’ diet mainly comprises fresh fish and other seafood.

It is possible that other influences may have a similareffect. Some foods, like garlic, have natural inhibitors of H.pylori growth (25). It is tempting to speculate that dietaryfactors (garlic, fresh fish) may determine a relatively higherprevalence of less virulent strains in low risk populations.

Two features of our study may have contributed to thepositive correlation found. In the first place, our sampleconsisted of volunteers who may better represent the generalpopulation than clinical series. The latter selects patients inwhom the virulence-related genotypes may be responsiblefor the disease, such as peptic ulcer or nonulcer dyspepsia.Clinical series may represent preselected infection outcomesrather than the general distribution of H. pylori genotypes inthe population. Previous reports from Africa and other pop-ulations at low risk for gastric cancer are mostly based onclinical (symptomatic) patients (26). Secondly, our studycompares populations in which markedly contrasting gastriccancer risks had been documented. This makes it moreprobable to detect minor differences in genotype distribu-tion between populations.

ACKNOWLEDGMENTS

Supported by grants P01-CA28842 from the National Can-cer Institute and HEF 2000 03-05 from the Louisiana Boardof Regents.

Reprint requests and correspondence: Pelayo Correa, M.D.,Department of Pathology, Louisiana State University Health Sci-ences Center, 1901 Perdido Street, New Orleans, LA 70112.

Received Feb. 20, 2002; accepted June 11, 2002.

REFERENCES

1. Holcombe C. Helicobacter pylori: The African enigma. Gut1990;33:429–31.

2. Blaser MJ. Helicobacter pylori and gastric diseases. BMJ1998;316:1507–10.

3. Covacci A, Telford J, Del Guidice G, et al. Helicobacter pylorivirulence and genetic geography. Science 1999;284:1328–33.

4. Hansson LE, Nyren O, Hsing A, et al. The risk of stomach

Table 2. H. pylori Genotypes in Subjects from Pasto andTumaco Regions of Colombia

Genotype Pasto Tumaco

CagA positive 160 60vacA

s1am1 7 1

s1bm1 142 48m2 3 1m1m2 2 2

s1a s1bm1 0 1m2 2 0m1m2 4 0

s1bs2

m1 0 2m2 0 2m1m2 0 3

CagA negative 17 14vacA

s1bm1 2 0m2 3 1

s2

m2 12 12s1bs2

m2 0 1

2841AJG – November, 2002 H. pylori Genotypes and the African Enigma

cancer in patients with gastric or duodenal ulcer disease.N Engl J Med 1996;335:242–8.

5. Uemura N, Okamoto S, Yamamoto S, et al. Helicobacterpylori infection and the development of gastric cancer. N EnglJ Med 2001:345:784–832.

6. Censini S, Lange C, Xiang Z, et al. Cag, a pathogenicity islandof Helicobacter pylori, encodes type I-specific and disease-associated virulence factors. Proc Natl Acad Sci U S A 1996;93:14648–53.

7. Atherton JC. The clinical relevance of strain types of Helico-bacter pylori. Gut 1997;40:701–3.

8. Atherton JC, Peek RM, Tham KT, et al. Clinical and patho-logical importance of heterogeneity in vac A, the vacuolatingcytotoxin gene of Helicobacter pylori. Gastroenterology 1997;112:92–5.

9. van Doorn LJ, Figueiredo C, Megraud F, et al. Geographicdistribution of vacA allelic types of Helicobacter pylori. Gas-troenterology 1999;116:823–30.

10. Atherton JC, Peek RM, Tham KT, et al. Clinical and patho-logical importance of heterogeneity in vacA, the vacuolatingcytotoxin gene of Helicobacter pylori. Gastroenterology 1997;112:92–5.

11. Correa P, Cuello C, Duque E. Carcinoma and intestinal meta-plasia of the stomach. J Natl Cancer Inst 1970;44:297–306.

12. Cuello C, Correa P, Haenszel W, et al. Gastric cancer inColombia. I. Cancer risk and suspect environmental agents.J Natl Cancer Inst 1976;57:1015–20.

13. Correa P, Cuello C, Duque E, et al. Patologıa geografica delcarcinoma y de la metaplasia intestinal del estomago en Co-lombia (Geographic pathology of carcinoma and intestinalmetaplasia of the stomach in Columbia). Antioquia Medica1969;56:41–56.

14. Cuello C, Correa P, Haenszel W, et al. Gastric cancer inColombia. III. Cancer risk and suspect environmental agents.J Natl Cancer Inst 1976;57:1027–35.

15. Llanos G, Correa P, Barbosa O. Morbilidad por cancer enCartagena Colombia (Morbidity from cancer in Cartagena,Colombia). Antioquia Medica 1969;19(5):377–88.

16. Correa P, Fox J, Fontham E, et al. Helicobacter pylori and

gastric carcinoma. Serum antibody prevalence in populationswith contrasting cancer risks. Cancer 1990;66:2569–74.

17. Garcia Del Risco F, Caraballo JM, Urbina D. Helicobacterpylori en sujetos asintomaticos, con ulcera duodenal, ulceragastrica y dispepsia no ulcerosa en Cartagena (Helicobacterpylori in asymptomatic subjects and in duodenal ulcer, gastriculcer, and nonulcer dyspepsia in Cartagena). Rev Col Gastro-enterol 1994;9:188–92.

18. Figueiredo C, van Doorn LJ, Nogueira JM, et al. Helicobacterpylori genotypes are associated with clinical outcome in Por-tuguese patients and show a high prevalence of infections withmultiple strains. Scand J Gastroenterol 2001:36:128–35.

19. Nogueira C, Figueiredo F, Carneiro A, et al. Helicobacterpylori genotypes may determine gastric histopathology. Am JPathol 2001:158:647–54.

20. van Doorn LJ, Figueiredo R, Rossa G, et al. Typing of theHelicobacter pylori vacA gene and detection of the cagA geneby PCR and reverse hybridization. J Clin Microbiol 1998;36:1271–6.

21. van Doorn LJ, Figueiredo C, Sanna R, et al. Clinical relevanceof the cagA, vacA, and iceA status of Helicobacter pylori.Gastroenterology 1998;115:58–66.

22. van Doorn LJ, Schneeberger PM, Nouhan N, et al. Importanceof Helicobacter pylori cagA and vacA status for the efficacy ofantibiotic treatment. Gut 2000;46:321–6.

23. El Omar E, Carrington M, Chow W, et al. Interleukin-1polymorphisms associated with increased risk of gastric can-cer. Nature 2000:404:398–402.

24. Correa P, van Doorn LJ, Bravo JC, et al. Unsuccessful treat-ment results in survival of less virulent genotypes of Helico-bacter pylori in Colombian patients. Am J Gastroenterol 2000;95:564–6.

25. Cellini L, Di Campli E, Masulli M, et al. Inhibition of Heli-cobacter pylori by garlic extract (Allium sativum). Federationof European Microbiological Societies Immunology and Med-ical Microbiology 1996;13:273–77.

26. Letley DP, Lastovica A, Louw JA, et al. Allelic diversity of theHelicobacter pylori vacuolating cytotoxin gene in SouthAfrica: Rarity of the vacA s1a genotype and natural occur-rence of an s2/m1 allele. J Clin Microbiol 1999;37:1203–5.

2842 Bravo et al. AJG – Vol. 97, No. 11, 2002