The recognition that Helicobacter pylori (Hp) plays an important role in the pathogenesis of several gastroduodenal disorders makes its diagnosis necessary in many clinical situations. Several invasive and non-invasive tests with variable sensitivity and specificity are employed for the purpose. Invasive tests like rapid urease test (RUT), histology and culture are possible only in centers with facilities for endoscopy and are often not preferred particularly in children. The non invasive tests (NITs) obviate the need for endoscopy and, in contrast to biopsy based methods which may fail to diagnose the infection due to patchy distribution of Hp, assess the global presence of Hp in the stomach even when the bacteria are irregularly distributed on the gastric mucosa. Interpretation of NIT, however, is limited since past infection and asymptomatic colonization may be tested as false positives by these methods. This study was undertaken to evaluate some of the NITs in comparison with conventionally used diagnostic methods, and particularly to assess the feasibility of using commercial serological kit for detection of IgG antibodies from saliva.

Children of either sex less than 12 years of age attending Division of Pediatric Gastro-enterology and Nutrition for various gastro-intestinal symptoms were prospectively investigated by upper gastrointestinal endo-scopy, rapid blood test, serum and salivary ELISA for IgG antibodies against Hp. Upper GI endoscopy (for biopsy, antral brushings, rapid urease test, and culture of the biopsy material) was performed with pediatric fibreoptic endoscope (GIF Type PQ 20, model Olympus) after written consent of the parents, adequate preparation and sedation of the child. Three biopsies from antrum and one from corpus were pinched with the help of biopsy forceps, one antral biopsy immediately used for RUT, another antral biopsy transferred to normal saline for impression smear and culture, and other two biopsies (antral and corpus) were preserved in 10% buffered formalin to be used for histopatho-logical examination. Brush cytology was performed with sheathed cytology brush with the endoscopic tip in the antrum and then brushing done with the mucosal surface on the inferior surface, laterally and on the superior surface of the brush, brush retracted under the sheath, taken out and vigorous to and fro brushing performed on glass slides, one smear air dried and processed for Giemsa staining and the other fixed with ethanol and processed for hematoxylin and eosin staining. For culture studies the biopsy specimen was transported in normal saline and processed within one hour. Crushed biopsy specimens were plated simultaneously on Skirrow’s medium and incubated in microaerophilic conditions (5-6% O2 , 8-10% CO2 , 80-85% N2 and 4-5% H2) provided with the help of ANOXOMAT (Martâ Microbiology BV). H. pylori was identified by colony characteristics, Gram’s stain and biochemical characteristics.

A blood sample was obtained by venepuncture and serum stored at –20ºC for Hp serology to detect specific IgG antibodies by serum ELISA (Varelisa H. pylori IgG anti-bodies enzyme immunoassay by Pharmacia and Up John by Biomerica) and Rapid Blood Test (EZ-Hp TM-a chromato-graphic absorbent membrane strip with immobilized Hp antigens). About 2 ml of unstimulated saliva was collected in plain vial stored at –20ºC for detection of IgG antibodies (Varelisa H. pylori IgG antibodies enzyme immunoassay by Pharmacia and Upjohn). Standardization of kits for detecting IgG antibodies for Hp was done by taking 20 controls. The controls were chosen randomly from children presenting in the Pediatrics OPD with non-gastroduodenal disorders and preferably age and sex matched with the study cases. The mean ODs of the control were measured and 2 SD above the mean was taken as the positive cut off value.

Criteria used in the study for diagnosis of Hp infection for comparison of various diagnostic methods was a minimum con-cordance of 3 out of 6 test parameters (histopathology of biopsy, brush cytology, impression smear, RUT, culture and serum ELISA). Wherever suitable, statistical methods like pearson chi-square test and fischer’s exact test (WHO epi-info software) were applied to find the significance of data obtained. Appropriate statistical methods were used to assess sensitivity (SE), speci-ficity (SP), positive predictive value (PPV) and negative predictive value (NPV) of these diagnostic modalities.

Thirty one children aged 3-12 years (mean age 7.4 years) including 17 boys (54.8%) and 14 girls (45.2%) with symptoms of chronic diarrhea (45.2%), recurrent abdominal pain (35.5%) and recurrent vomiting (19.3%) were subjected to various investigations for Helicobacter pylori (Hp) infection. Six (19.3%) cases fulfilled the diagnostic criteria for Hp infection as laid down for the study i.e., concordance of 3 or more of the test parameters. In 5 (16.1%) cases there was concordance of 4 or more parameters and in 3 (9.6%) cases concordance occurred for 5 more parameters. Overall RUT was positive in 8 (25.6%), serum ELISA in 7 (22.5%), salivary ELISA and brush cytology in 5 each (16.1%), and Gram’s impression smear, culture and histology in 4 (12.9%) cases each. Rapid blood test was positive in 3 (9.7%) cases. Ten per cent of controls also tested positive with serum ELISA.

Considering 6 cases as positive for Hp as per the diagnostic criteria used in the study, serology was positive for Hp in all (100%), brush cytology and RUT in 5 (83.3%) cases each, and Gram’s impression smear, histology and culture in only 4 (66.7%) cases each. Sensitivity, specificity, positive predictive value and negative predictive value of various diagnostic methods is provided in Table 1. When compared with serum ELISA, the results of salivary ELISA were better than rapid blood test. Duration of incubation for RUT increased the sensitivity from ½ hour to 24 hours but specificity dropped after 4 hours incubation (Table I). Hematoxylin and eosin stain had higher positivity with brush cytology as compared with histology.

Table I__ Sensitivity, Specificity, Positive Predictive Value and Negative Predictive Value of Various Diagnostic Methods in Comparison with cases (n = 6) Diagnosed to have H. pylori infection

Methods

Sensitivity

Specificity

PPV

NPV

Rapid urease test

½ hr

16.7

100

100.0

83.3

1 hr

66.7

100

100.0

92.6

4 hr

83.3

100

100.0

96.1

24 hrs

83.3

88

62.5

95.6

Serum ELISA

100.0

96

85.7

100.0

Salivary ELISA

66.7

96

80.0

92.3

Brush cytology

83.3

100

100.0

96.1

Gram’s impression smear

66.7

100

100.0

92.6

Culture

66.7

100

100.0

92.6

Histology

66.7

100

100.0

92.6

Rapid blood test

50.0

100

100.0

89.3

PPV = Positive predictive value; NPV = Negative predictive value.

Detection of Hp infection by several earlier studies conducted in the same pediatric population has been reported from 14-16%(1). In view of the limitations of each diagnostic test we did not consider any of the available tests as ‘gold standard’. However, for comparison of results we used concordance of 3 or more positive tests, out of a total of 6 conventionally used tests, to consider it as true positive. This diagnostic criteria enabled us to diagnose Hp infection in 6/31 (19.3%) cases, which was lower than results of individual tests like RUT (25.6%) and serum ELISA (22.5%) but higher than culture and histology (12.9%). Our observations are in agreement with the results of previous studies suggesting that the prevalence of Hp infection in our pediatric population is higher than developed countries but not as high as reported by some clinical and epidemiological surveys(2).

Out of the 6 diagnostic methods commonly utilized in clinical practice, considering their positioning on the ROC curve, serum ELISA, RUT and brush cytology had a reasonably good sensitivity and specificity. Gram’s impression smear test, culture and histology were the least sensitive but with excellent specificity. RUT tended to increase in sensitivity if incubation period was increased beyond ½ an hour but dropped in its specificity beyond 4 hours incubation. Marshal et al(3) observed that positivity increased to 98% at 24 hours but in our cases the positivity did not increase any further if incubation was done for 24 hours. Therefore we concluded that the optimum duration of incubation for RUT for a good sensitivity and 100% specificity is 4 hours. The false positivity of RUT was 8% (2/25) which could be attributed to presence of Gastrospirillum hominis and other contaminating bacteria such as Pseudomonas and Proteus which are known to test false positives(4). In 1 of the 6 cases RUT was negative which can be explained due to patchy distribution of the organism and due to inactivity of urease enzyme with the use of anti ulcer drugs(5).

Brush cytology was observed to be better than histology because in brush cytology the organisms were identified and distinguished by their characteristic ‘spiral’, ‘cork screw’ or ‘seagul’ shaped morphology lying in amorphous mucinous background. Similar characterstics morphology was not as distinct on histology. Moreover, brush cytology screens a much larger area of antral mucosa than a grasp biopsy. Efficacy of brush cytology as an adjunct to grasp biopsy has also been suggested in our earlier reports.

Isolation of Hp on culture has been reported to be quite low ranging from 20-30%, but the yield was reported to be as high as 89% when antral and fundic biopsy was combined(6). Our results suggest a fairly good sensitivity (66.7%) with high specificity (100%) of culture. Gram’s impression smear had a lower sensitivity but 100% specificity and was comparable with histology and culture. These observations are in close agreement with other studies(6,7).

Estimation of Hp specific IgG antibodies using enzyme immunoassay by serum ELISA has been a commonly used non-invasive investigation for screening and field surveys. Our results with serum ELISA are in agreement with earlier published reports(8,9). However, variation in the results of various studies may reflect the presence of several serotypes of Hp. Moreover, 10% of the control group also showed positive results which could be attributed to either past infection or asymptomatic colonization. The interest in rapid diagnosis of Hp has increased in recent years. The rapid blood test has been designed to fulfil this need but it had the lowest sensitivity of 50% (and a specificity of 100%) which is lower as compared to preliminary results of other studies with a sensitivity and specificity ranging from 63-91% and 83-94% respectively(10,11).

Salivary ELISA was detected in 5 (16.1%) cases as against 7(22.6%) cases by serum ELISA. In comparison to diagnostic criteria laid down for the study, salivary ELISA rated lower than serum ELISA but compared very well with its specificity as it was tested negative in all those cases tested negative by serum ELISA. Sensitivity and specificity of salivary ELISA has been reported to range from 64-93% and 58-95% respectively by several studies(12-15). Salivary PCR for Hp has also reported to have low sensitivity(16). Several factors may contribute to lower positivity with salivary ELISA as compared to others(17) and discrepancy in the positive results between serum and salivary ELISA including immunoglobulin degradation by salivary proteases.

Our results suggest that for detection of Hp, amongst various invasive methods, brush cytology was the most sensitive and specific. RUT also had good sensitivity but lower specificity. Gram’s impression smear, histology and culture were very specific but there is a possibility of missing the diagnosis in 33.3% cases. In non-invasive methods serum ELISA could be considered as the investigation of choice because of high sensitivity and a reasonable specificity. Being non-invasive, serum ELISA can also be used for sero-epidemiological surveys. Rapid blood test appears to be handicapped with a low sensitivity. Salivary Hp IgG assay can not be regarded as a first choice procedure to screen for Hp but can be useful when serum samples are not available or difficult to get and other tests are not possible particularly in young children.

Contributors: NG participated in data collection, performing microbiological tests and in analysis. BLS designed the study and helped in interpretation of results and preparation of manuscript. AKP helped in designing the study, performing endoscopic procedures and in preparation of manuscript. He will act as the guarantor for the article. PB and MC participated in cnducting pathological investigations and critical review of the manuscript.

Funding: None.

Competing interests: None stated.