Helicobacter pylori causes chronic inflammation of the gastric mucosa and has been identified in tracheobronchial secretions. Serum IgG against H. pylori was therefore measured prospectively in consecutive subjects with bronchiectasis (n = 100; mean age ± SD 55.1 ± 16.7 yr), active pulmonary tuberculosis (n = 87; age, 57.3 ± 19.1 yr), and healthy volunteers (n = 94; age, 54.6 ± 7.6 yr). Seropositivity was found in 76.0% of bronchiectatic subjects, which was significantly higher than that of the control (54.3%, p = 0.001) and tuberculous (52.9%, p = 0.0001) groups. Multiple logistic regression, adjusted for age, sex, occupational social class, and number of persons living in the household, showed that H. pylori IgG levels of the bronchiectatic group were still significantly higher than that of the control (p = 0.0014) and tuberculous (p = 0.0154) groups. Multiple regression analysis revealed associations between H. pylori serology and sputum volume (p = 0.03) and age (p = 0.001) in the bronchiectatic patients, but not lung function indices or causes of bronchiectasis. The H. pylori seroprevalence in bronchiectasis was significantly (p = 0.0002) higher in patients who produced more (83.1%) than those who produced less than 5 ml sputum/24 h (58.6%). This is the first report of a high H. pylori seroprevalence in bronchiectasis which appears to be specific. Further studies are indicated to evaluate the possible pathogenic role of H. pylori in bronchiectasis.
The relatively recent rediscovery of Helicobacter pylori has revolutionized insights into the pathogenesis and treatment of peptic ulcer disease (1). H. pylori is a microaerophilic gram-negative spiral-shaped bacterium which is causally related to chronic active gastritis (2), peptic ulcer disease (2), primary low-grade B-cell gastric lymphoma (3), and gastric carcinoma (4, 5). Epidemiological studies have revealed an association of H. pylori seroprevalence with increasing age (6), lower socioeconomic status (6, 7), and crowdedness in the household (8) but not gastroesophageal reflux (9) or sex (10). Recently, a high H. pylori seroprevalence has also been found in ischemic heart disease (11), rosacea (12), and childhood growth retardation (7).
Bronchiectasis is a chronic infective and inflammatory disease of the tracheobronchial tree and affected patients suffer from recurrent sputum production, hemoptysis, and exacerbations. Although many known causes of bronchiectasis have been identified, over 60% of cases are regarded as idiopathic (13). Despite the disappearance of the original causative assault to the respiratory tract, such as pertussis, these patients continue to produce significant amounts of sputum which indicates an underlying active tracheobronchial inflammation (14). H. pylori has been recently identified in the tracheobronchial aspirates in mechanically ventilated patients and the possibility that it might cause ventilator-associated pneumonia has been raised (15). In addition, other bacteria found in the gastric juice have been isolated from the respiratory tract in situations which favored bacterial colonization of the stomach (16) and where regurgitation of gastric contents into the respiratory tract can occur (17). Although H. pylori is found in tracheobronchial secretions and its role in gastric mucosal inflammation is similar to the pathogenesis in bronchiectasis, i.e., cytokine-mediated (18-21), the seroprevalence of H. pylori in bronchiectasis has not been studied previously.
Three groups of subjects, who consented to the provision of venous blood, were recruited consecutively (from January 1996 to December 1996). One hundred patients who suffered from bronchiectasis (diagnosed by typical clinical symptoms and high-resolution computed tomography) who were in steady state (defined by no significant changes in respiratory symptoms or signs for at least 3 wk) and had no active tuberculosis were recruited. Consecutive patients with active pulmonary tuberculosis (diagnosed by sputum microbiology and/or thoracic radiology) were recruited if they were still receiving antituberculous chemotherapy and did not suffer from bronchiectasis. Blood was also collected from normal subjects who attended health exhibitions designed for public health education who themselves had no known history of peptic ulcer, cerebrovascular, ischemic heart, or respiratory diseases. All the subjects were questioned on the number of persons living in the same household and occupationally classified according to the British Registrar General, as classes 1, 2, 3, 4, and 5 for professional, intermediate, skilled manual, partly skilled manual, and unskilled manual workers respectively. The procedures had approval from the institutional ethics committee on human research.
The following data were collected from the bronchiectatic patients: age, sex, spirometry, i.e., forced expiratory volume in one second (FEV1) and forced vital capacity (FVC) (in percent predicted), 24-h sputum volume collected at home (0–5, 5–10, 10–15, 15–20, 20–30, 30– 40, 40–50, 50–100 ml/24 h as grades 1, 2, 3, 4, 5, 6, 7, 8, respectively), smoking history, history of known ischemic heart disease and proven peptic ulcer, other medical conditions, and the etiology of bronchiectasis (13).
Aliquots of 100 μl of IgG calibrator, control and diluted serum from patient or control groups (stored at −70° C before use) were pipetted separately into the microliter wells (coated with purified H. pylori group specific antigens) of a commercially available ELISA kit (Bio-Rad, Hercules, CA) and discarded after incubation for 60 min at 25° C. The wells were washed three times with buffer solution and blotted dry before addition of 100 μl of anti-IgG enzyme conjugate and incubation at 25° C for 30 min. After washing, the addition of another 100 μl of the working solution was followed by incubation at 25° C for 10 min, after which 50 μl of stop solution was added at a rapid steady pace into each of the microwells. The absorbance of the wells at 450 nm was read within 20 min and the optical density (and its corresponding concentration) was determined for each batch of experiments. A positive, equivocal, and negative result for IgG against H. pylori was assigned when the concentration was > 20, between 12.5 and 20, and < 12.5 units (U)/ml, respectively. By using histological examination (Warthin-Starry stain) as the gold standard (22), the sensitivity and specificity of this ELISA kit in our patient population have been found to be 95 and 85% respectively. In the event of an equivocal result, a repeat test was performed on serum obtained 2 wk after the initial venesection, as suggested by the manufacturer.
Serum total IgG level was determined using routine nephelometry at the Clinical Biochemistry Department of the University of Hong Kong. The ratio of H. pylori–specific IgG to total serum IgG was determined for each patient. Serum IgG levels against the viral capsid antigen (VCA) of Epstein-Barr virus was determined by using routine established methodology at the Clinical Microbiology Laboratory of the University of Hong Kong. Briefly, IgG against Epstein-Barr VCA was determined by immunofluorescent technique using fluorescein isothiocyanate–conjugated and heavy chain–specific goat anti-human sera (Dako, Glostrup, Denmark). Titers were expressed as the reciprocal of the maximal dilution that gave a positive immunofluorescence as described previously (23). Freshly obtained sputum was plated on Columbia Blood Agar (Oxoid, Basingstoke, UK) with H. pylori Selective Supplements (Oxoid, Basingstoke, UK) which contained vancomycin (10 mg/L), trimethoprim (5 mg/L), cefulodin (5 mg/L), and amphotericin B (5 mg/L). Inoculation was performed within 30 min of sputum collection under standard conditions using Oxoid AnaeroJars and CampyGen (Oxoid, Basingstoke, UK) for at least 3 d. Gram staining and standard microbiological identification procedures were performed on all colonies isolated (24).
The statistical analysis comprised the Kruskal-Wallis K sample test for comparison of the central tendency between the three groups, i.e., the nonparametric one-way analysis of variance; pair-testing between two groups was made in terms of the Mann-Whitney–Wilcoxon test. A nonparametric test was applied because of positively skewed H. pylori serology raw values; a logarithmic transformation gave a distribution acceptably close to normal samples, but this meant that “zero” serology values could not be used. Fisher exact test was used for binomial data, and the results were then given in terms of odds ratios. A logistic multiple regression model was applied to the data; the serum H. pylori IgG status (positive or otherwise) constituted the dependent variable, and sex, age, number of persons living in the same household, and occupational social class were the independent variables together with the patient group variable. A multiple linear regression was applied to the bronchiectatic patient data; the dependent measure was the serum H. pylori IgG concentration, and the sputum volume and age were the independent measures. A p value below 0.05 was regarded as significant except for the pair-testing situation, when a p value below 0.01 was used. The analysis was performed in the SAS statistical analysis system (25).
There was no significant difference between the age, number of persons living in the same household, and occupational social class among the three groups (p = 0.1198) (Table 1). For the bronchiectatic group, the FEV1 and FVC, 24-h sputum grading, and percent of patients with known ischemic heart disease are shown in Table 2. None of the subjects suffered from rosacea. The etiology of bronchiectasis was considered to be idiopathic, post-tuberculous, postpneumonic, Kartagener's syndrome, or diffuse panbronchiolitis (Table 2). Four bronchiectatic patients, but none of the tuberculous or healthy subjects, had proven peptic ulcer disease. Of these, three patients were seropositive for H. pylori. There was no significant difference between the two subgroups of bronchiectatic patients who had a sputum volume grading of 1 or > 1 in the aforementioned parameters.
|Controln = 94 (M = 32)||Bronchiectasis (All)n = 100 (38)||Bronchiectasis(24-h sputum > 5 ml)|
n = 71 (26)
|Bronchiectasis(24-h sputum = 0–5 ml)|
n = 29 (12)
|Tuberculosisn = 87 (71)|
|Mean ± SD||54.6 ± 7.55||55.1 ± 16.67||55.2 ± 17.19||54.7 ± 15.62||57.3 ± 19.10|
|95% confidence interval||53.1–56.1||51.8–58.4||51.2–59.2||49.0–60.4||53.3–61.3|
|HP seropositivity* (% of patients)||54.3||76||83.1||58.6||52.9|
|Serum HP IgG concentration (U/ml)|
|Mean ± SD, median||29.1 ± 23.8, 24.4||41.0 ± 26.02, 41.7||46.0 ± 25.11, 47.7||28.9 ± 24.61, 31.0||24.7 ± 17.53, 21.7|
|95% confidence interval||24.3–33.9||35.9–46.1||40.2–51.8||19.9–37.9||21.0–28|
|Occupational social class (1-5)†|
|Mean ± SD||3.8 ± 0.95||3.8 ± 0.98||3.8 ± 0.97||3.8 ± 0.99||3.7 ± 0.94|
|95% confidence interval||3.6–4.0||3.6–4.0||3.6–4.0||3.4–4.2||3.5–3.9|
|Number of persons living in the same household|
|Mean ± SD||3.9 ± 1.54||3.7 ± 1.62||3.7 ± 1.63||3.8 ± 1.63||4.4 ± 2.16|
|95% confidence interval||3.6–4.2||3.4–4.0||3.3–4.1||3.2–4.4||3.9–4.9|
|Parameter||Bronchiectasis (All)(n = 100)||Bronchiectasis (24-h sputum = 0–5 ml)(n = 29)||Bronchiectasis (24-h sputum > 5 ml)(n = 71)|
|FEV1, % pred|
|Mean ± SD, Median||66.6 ± 29.59, 67||73.4 ± 28.92, 81||63.8 ± 3.51, 63|
|Range; 5th and 95th centiles||15–127; 60.8, 72.5||15–113; 62.4, 84.5||15–127; 56.8, 78.6|
|FVC, % pred|
|Mean ± SD, median||74.5 ± 26.13, 73||79.5 ± 27.37, 84||72.5 ± 3.03, 70|
|Range; 5th and 95th centiles||15–127; 69.3, 79.7||37–129; 69.1, 89.9||25–136; 66.5, 78.6|
|Daily sputum volume grading, Mean ± SD||3.4 ± 0.23||1 ± 0||4.4 ± 0.25|
|Known ischemic heart disease, %*||1||3.4||0|
|Etiology of bronchiectasis, %*|
The mean level of H. pylori IgG for the bronchiectatic group was significantly higher than that of the control (p = 0.001) and tuberculous (p = 0.001) groups (Table 1). Using the manufacturer recommended cutoff points, 76.0% of cases were definitely positive, which was significantly higher than control (54.3%, p = 0.001) and tuberculous groups (52.9%, p = 0.0001). There was no significant difference in the IgG levels or the seropositivity between the control and tuberculous groups (p = 0.452). H. pylori IgG levels, using the continuous scale, were compared among different groups by using the Kruskal-Wallis test, and the differences were significant (p = 0.0001).
There was no significant difference (p = 0.569) in the serology status between the control and the tuberculous subjects when adjusted for the different possible confounding factors (odds ratio [OR] = 0.82, 95% confidence interval [CI] 0.41 to 1.63) (Table 3). However, a significant difference was found between the bronchiectatic and the control (p = 0.0014) and the tuberculous groups (p = 0.0154). The OR was 2.81 (95% CI 1.49 to 5.29) for a positive H. pylori serology status in the bronchiectatic group when compared with the control group and 2.39 (95% CI 1.18 to 4.84) when compared with tuberculous patients. Within the bronchiectasis group, the idiopathic subgroup had a seroprevalence of 75.6% which was not significantly different from the subgroup which had an identifiable etiology (77.8%) for bronchiectasis (p > 0.05).
|Control versus Bronchiectasis‡||Tuberculosis versus Bronchiectasis‡||Control versus Tuberculosis‡|
|Odds Ratio||95% CI||p Value||Odds Ratio||95% CI||p Value||Odds Ratio||95% CI||p Value|
|No. of persons|
|in the same|
A multiple regression analysis was applied to the H. pylori–specific IgG values of the series of 100 patients, which were expressed in a continuous scale, as the dependent measure. The following measures were kept in the final model: sputum volume (t = 2.1, p = 0.03), age (t = 3.2, p = 0.001), and female sex (t = 2.1, p = 0.04). The R-square (the squared multiple correlation coefficient) was 0.17 and the F value for the whole regression model was 6.7 (p = 0.0001). Other variables included in the model that were not significant were: smoking, FEV1, and FVC. Increased H. pylori–specific IgG values were associated with increased age, higher sputum volume, and female sex. Using a logistic multiple regression analysis, with a H. pylori IgG concentration of > 20 U/ml as the cutoff point and adjustment made for age, number of persons living in the same household, occupational social class, and sputum production, the H. pylori seroprevalence in female (80.6%) was not significantly different from that of the male (65.8%) bronchiectatic patients (p > 0.05).
The median (range) total serum IgG levels were 1,410 (356 to 2,428), 1,480 (356 to 2,428), and 1,270 (804 to 2,130) mg/dl for the entire (n = 100), H. pylori seropositive (n = 76), and H. pylori seronegative (n = 24) groups respectively. There was a significant difference between the total serum IgG in the H. pylori seropositive and the H. pylori seronegative groups (p = 0.019). The median (range) ratios of H. pylori–specific IgG to total serum IgG levels for the entire, H. pylori seropositive and the H. pylori seronegative groups were 0.029 (0.001 to 0.080), 0.037 (0.009 to 0.080), and 0.004 (0.001 to 0.011) (U × dl/ml × mg), respectively. The latter two groups were also significantly different (p = 0.001). Only one patient, a 45-yr-old female with H. pylori–specific IgG of 71.3 U/ml, had undetectable IgG against the VCA of Epstein-Barr virus among the 100 patients with bronchiectasis. The median (range) IgG against Epstein-Barr VCA in the entire, H. pylori seropositive, and H. pylori seronegative groups of bronchiectasis patients were 1/1,280 (1/10,240 to 1/80), 1/1,280 (1/2,415 to 1/160), and 1/1,280 (1/10,240 to 1/80), respectively. There was no significant difference in IgG against Epstein-Barr VCA between the latter two subgroups (p = 0.76). There was no correlation between the levels of H. pylori–specific IgG with IgG against Epstein-Barr virus (Spearman rank r = 0.04, p = 0.67). H. pylori was not isolated in the sputum of the 30 patients in the pilot study.
This study shows that there is a high seroprevalence of H. pylori infection in bronchiectasis (76%) which is significantly higher than that of the normal volunteers (54.3%) and tuberculous patients (52.9%). The “crowdedness” (number of persons living in the same household), occupational social class, and age, which are known to be associated with H. pylori seroprevalence (6, 7), were similar in the three groups of subjects. It is very likely that the abnormally high seroprevalence is specific to bronchiectasis as there was no association with tuberculosis, another chronic infective and inflammatory lung condition. Among the bronchiectatic patients, the sputum producers (i.e., those with 24-h sputum grading > 1) had a H. pylori seroprevalence significantly higher (83.1%) than that of the nonproducers (58.6%) (p = 0.0002). The latter was not significantly different from the control group, which suggests that H. pylori seroprevalence is also specific to the sputum producers (i.e., those with active underlying tracheobronchial inflammation) irrespective of the etiology (idiopathic or not) of bronchiectasis. Multiple regression analysis applied to the data of bronchiectatic patients revealed associations between H. pylori IgG concentrations (continuous values) and sputum volume, age, and female sex, although the sex effect did not remain significant after logistic regression analysis. The other parameters in the regression model, including lung spirometry and etiology of bronchiectasis, showed no correlation with H. pylori serology. As we only employed meticulous history taking and clinical chart review to disclose known diagnosis of peptic ulcer disease, it is possible that, without restoring to gastroscopy and gastric biopsies, peptic ulcer disease might have been missed in some of our bronchiectatic, tuberculous, and healthy subjects. However, it is very unlikely that the significantly higher seroprevalence in bronchiectasis could be entirely accounted for by the missed “silent peptic ulcers” in our bronchiectatic subjects.
A higher serum total IgG was found in the H. pylori seropositive than the seronegative subgroup of patients with bronchiectasis. This raises the possibility that the increase in H. pylori–specific IgG might be part of a hyperimmune state in active bronchiectasis rather than a specific phenomenon (13, 14, 18, 20). However, this is extremely unlikely as the ratio of H. pylori–specific IgG to total serum IgG was still significantly higher in the H. pylori seropositive compared with the seronegative group of bronchiectatic patients (p = 0.001). In addition, there was no difference in the levels of Epstein-Barr VCA-specific IgG between the H. pylori seropositive and seronegative subgroups of patients with bronchiectasis (p = 0.76). Finally this possible “polyclonal response” as a cause of the high concentrations of H. pylori IgG among the sputum producers is not supported by the lack of correlation between the levels of IgG against H. pylori and Epstein-Barr VCA (r = 0.04, p = 0.67). Our findings are therefore strongly indicative of a specific association of H. pylori seroprevalence with disease activity in bronchiectasis.
There are interesting similarities between the pathogenesis of bronchiectasis and ulcerogenesis. In both conditions, there is extensive recruitment of polymorphs and T lymphocytes into the submucosa (18, 19) and cytokine release especially of interleukin-8 (IL-8), tumor necrosis factor-α, and IL-1 (20, 21). Respiratory pathogens such as Haemophilus influenzae and Pseudomonas aeruginosa persist in the tracheobronchial tree (26), similar to H. pylori persisting in the stomach for years to decades (27), and these organisms also cause similar ultrastructural damage to the target mucosal cells, such as vacuolation (28-30). Similar to the adherence of Haemophilus influenzae and P. aeruginosa to respiratory mucosa (31), H. pylori also appears to use silica acid residues as its putative receptor on gastric mucosa (32) and has high affinity for mucus (29, 30, 33).
Despite the above similarities in the pathogenesis of bronchiectasis and peptic ulcer disease, and the high seroprevalence of H. pylori in bronchiectasis, the possible pathogenic role of H. pylori in bronchiectasis remains unexplored. Seroprevalence of H. pylori, but not bronchiectasis, is associated with lower socioeconomic status (6) and occurs in clusters (8). This suggests that even if H. pylori has a pathogenic role in bronchiectasis, other confounding factors need to be present. The similarities between the pathogenesis in bronchiectasis and ulcerogenesis might also be a reflection of chronic inflammatory disease or bacteria-induced tissue damage. It is also possible that spilling or inhalation of H. pylori or their exotoxins into the respiratory tract might occur in H. pylori infection of the gastrointestinal tract, particularly in light of the high incidence of gastroesophageal reflux in bronchiectasis (34).
To our knowledge, this is the first report of a high seroprevalence of H. pylori specific to active bronchiectasis irrespective of its etiology. Although the evidence of an association between the seroprevalence of H. pylori and bronchiectasis appears strong, we were unable to isolate H. pylori in the sputum of bronchiectatic patients. To our knowledge, H. pylori has not been isolated in sputum although it has been identified by using Gram film examination of tracheal aspirate in mechanically ventilated patients in the intensive care setting (15). Further studies designed to evaluate the effects of H. pylori on the respiratory tract and to identify H. pylori in the bronchiectatic respiratory tracts, and clinical longitudinal studies to evaluate the effects of treatment of H. pylori in bronchiectasis should follow this study.
The authors thank all the subjects who donated their blood. Dr. P. L. Ho of the Department of Clinical Microbiology, the University of Hong Kong, provided expert microbiological opinion. Eileen Kwok, Raymond Leung, and C. S. Ho provided excellent technical assistance in this study.
Supported by a Peptic Ulcer Research Grant (The University of Hong Kong).
|1.||Graham D. Y.Treatment of peptic ulcers caused by Helicobacter pylori. N. Engl. J. Med.3281993349350|
|2.||Marshall J., Warren J. R.Unidentified curved bacilli in the stomach of patients with gastritis and peptic ulceration. Lanceti198413111314|
|3.||Isaacson P. G., Spencer J.Is gastric lymphoma an infectious disease? Hum. Pathol.241993569570|
|4.||Parsonnet J., Friedman G. D., Vandersteen D. P., Chang Y., Vogelman J. H., Orentreich N., Sibley R. K.Helicobacter pylori infection and the risk of gastric carcinoma. N. Engl. J. Med.325199111271131|
|5.||The EUROGAST Study GroupAn international association between Helicobacter pylori infection and gastric cancer. Lancet341199313591362|
|6.||Graham D. Y., Malaty H. M., Evans D. G., Evans D. J., Klein P. D., Adam E.Epidemiology of Helicobacter pylori in an asymptomatic population in the United States: effects of age, race, and socioeconomic status. Gastroenterol100199114951501|
|7.||Patel P., Mendall M. A., Khulusi S., Northfield T. C., Strachan D. P.Helicobacter pylori infection in childhood: risk factors and effect on growth. B.M.J.309199411191123|
|8.||Mendall M. A., Goggin P. M., Molineaux N., Levy J., Toosy T., Strachan D., Northfield T. C.Childhood living conditions and Helicobacter pylori seropositivity in adult life. Lancet3391992896897|
|9.||O'Connor H. J., Cunnane K.Helicobacter pylori and gastro-oesophageal reflux disease—a prospective study. Ir. J. Med. Sci.1631994369373|
|10.||The EUROGAST Study GroupEpidemiology of, and risk factors for, Helicobacter pylori infection among 3194 asymptomatic subjects in 17 populations. Gut34199316721676|
|11.||Mendall M. A., Goggin P. M., Molineaux N., Levy J., Toosy T., Strachan D., Northfield T. C.Relation of Helicobacter pylori infection and coronary heart disease. Br. Heart J.711994437439|
|12.||Rebora A., Drago F., Picciotto A.Helicobacter pylori in patients with rosacea. Am. J. Gastroenterol.89199416031604|
|13.||Cole, P. J. 1995. Bronchiectasis. In R. A. L. Brewis, B. Corrin, D. M. Geddes, and G. J. Gibson, editors. Respiratory Medicine. Saunders, London. 1995:1286–1317.|
|14.||Cole P. J.Inflammation: a two edged-sword—the model of bronchiectasis. Eur. J. Respir. Dis. Suppl.1471986615|
|15.||Mitz H. S., Farber S. S.Demonstration of Helicobacter pylori in tracheal secretions. J. Am. Osteopath. Assoc.9319938791|
|16.||Du Moulin G. C., Paterson D. G., Hedley-Whyte J., Lisbon A.Aspiration of gastric bacteria in antacid-treated patients. Lanceti1982242245|
|17.||Ibanez, J., A. Penafiel, J. M. Raurich, P. Marse, J. C. Paternostro, and F. Mata. 1988. Gastrooesophageal reflux and aspiration of gastric contents during nasogastric feeding, the effects of posture (abstract). Intens. Care Med. 14(Suppl. 2):296.|
|18.||Lapa e Silva J. R., Jones J. A. H., Noble B., Cole P. J., Poulter L. W.The immunological component of the cellular inflammatory infiltrate in bronchiectasis. Thorax441989668673|
|19.||Rauws E. A., Langenberg W., Houthoff H. J., Zanen H. C., Tytgat G. N.Campylobacter pyloridis–associated chronic active antral gastritis. Gastroenterol.9419883340|
|20.||Eller J., Lapa e Silva J. R., Poulter L. W., Lode H., Cole P. J.Cells and cytokines in chronic bronchial infection. Ann. NY Acad. Sci.7251994331345|
|21.||Noach L. A., Bosma N. B., Jansen J., Hoek F. J., van Deventer S. J., Tytgat G. N.Mucosal tumour necrotic factor-α, interleukin-1β, and interleukin 8 production in patients with Helicobacter pylori infection. Scand. J. Gastroenterol.291994425429|
|22.||Cutler A. F.Testing for Helicobacter pylori in clinical practice. Am. J. Med.100199635S41S|
|23.||Ho H. C., Ng M. H., Kwan K. C.Factors affecting serum IgA antibody to Epstein Barr viral capsid antigens in nasopharyngeal carcinoma. Br. J. Cancer371978356362|
|24.||Glupczynski, Y. 1996. Culture of Helicobacter pylori from gastric biopsies and antimicrobial susceptibility testing. In A. Lee and F. Megraud, editors. Helicobacter pylori: Techniques for Clinical Diagnosis and Basic Research. Saunders, London. 17–32.|
|25.||SAS Institute Inc. 1995. SAS user's guide (statistics version), 5th ed. SAS Institute, Cary, NC.|
|26.||Groeneveld K., van Alphen L., Eijk P. P., Visscher G., Jansen H. M., Zanen H. C.Endogenous and exogenous reinfections by Haemophilus influenzae in patients with chronic obstructive pulmonary disease: the effects of antibiotic treatment on persistence. J. Infect. Dis.1611990512517|
|27.||Blaser M. J.Helicobacter pylori and the pathogenesis of gastroduodenal inflammation. J. Infect. Dis.1611990626633|
|28.||Tsang K. W., Rutman A., Kanthakumar K., Belcher J., Lund V., Roberts D. E., Cole P. J., Wilson R.Haemophilus influenzae infection of human respiratory mucosa in low concentrations of antibiotics. Am. Rev. Respir. Dis.1481993201207|
|29.||Tsang K. W., Rutman A., Tanaka E., Lund V., Dewer A., Cole P. J., Wilson R.Interaction of Pseudomonas aeruginosa with human respiratory mucosa in vitro. Eur. Respir. J.7199417461753|
|30.||Leunk R. D., Johnson P. T., David B. C., Kraft W. G., Morgan D. R.Cytotoxic activity in broth-culture filtrates of Campylobacter pylori. J. Med. Microbiol.2619889399|
|31.||Ramphal R., Pyle M.Further characterization of the tracheal receptor for Pseudomonas aeruginosa. Eur. J. Clin. Microbiol.41985160162|
|32.||Evans D. G., Karjalainen T. K., Evans D. J., Graham D. Y., Lee C. H.Cloning, nucleotide sequence, and expression of a gene encoding an adhesin subunit protein of Helicobacter pylori. J. Bacteriol.1751993674683|
|33.||Hemalatha S. G., Drumm B., Sherman P.Adherence of Helicobacter pylori to human gastric epithelial cells in vitro. J. Med. Microbiol.351991197202|
|34.||Tsang K. W., Hu W., Lam W. K., Ip M., Lam S. K.A preliminary study of gastrooesophageal reflux in bronchiectasis (abstract). Am. J. Respir. Crit. Care Med.1551997A107|